Home/Federal/Federal Register/2023-16377

Proposed Rule2023-16377

Conflicts of Interest Associated With the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers

Primary source

Metadata and text below are from the Federal Register, a public-domain U.S. government work. Always verify the official published version before relying on it for any legal matter.

Published

August 9, 2023

Issuing agencies

Securities and Exchange Commission

Abstract

The Securities and Exchange Commission ("Commission" or "SEC") is proposing new rules ("proposed conflicts rules") under the Securities Exchange Act of 1934 ("Exchange Act") and the Investment Advisers Act of 1940 ("Advisers Act") to eliminate, or neutralize the effect of, certain conflicts of interest associated with broker-dealers' or investment advisers' interactions with investors through these firms' use of technologies that optimize for, predict, guide, forecast, or direct investment-related behaviors or outcomes. The Commission is also proposing amendments to rules under the Exchange Act and Advisers Act that would require firms to make and maintain certain records in accordance with the proposed conflicts rules.

Full Text

<html>
<head>
<title>Federal Register, Volume 88 Issue 152 (Wednesday, August 9, 2023)</title>
</head>
<body><pre>
[Federal Register Volume 88, Number 152 (Wednesday, August 9, 2023)]
[Proposed Rules]
[Pages 53960-54024]
From the Federal Register Online via the Government Publishing Office [<a href="http://www.gpo.gov">www.gpo.gov</a>]
[FR Doc No: 2023-16377]

[[Page 53959]]

Vol. 88

Wednesday,

No. 152

August 9, 2023

Part II

Securities and Exchange Commission

-----------------------------------------------------------------------

17 CFR Parts 240 and 275

Conflicts of Interest Associated With the Use of Predictive Data
Analytics by Broker-Dealers and Investment Advisers; Proposed Rule

Federal Register / Vol. 88 , No. 152 / Wednesday, August 9, 2023 /
Proposed Rules

[[Page 53960]]

-----------------------------------------------------------------------

SECURITIES AND EXCHANGE COMMISSION

17 CFR Parts 240 and 275

[Release Nos. 34-97990; IA-6353; File No. S7-12-23]
RIN 3235-AN00; 3235-AN14

Conflicts of Interest Associated With the Use of Predictive Data
Analytics by Broker-Dealers and Investment Advisers

AGENCY: Securities and Exchange Commission.

ACTION: Proposed rule.

-----------------------------------------------------------------------

SUMMARY: The Securities and Exchange Commission (``Commission'' or
``SEC'') is proposing new rules (``proposed conflicts rules'') under
the Securities Exchange Act of 1934 (``Exchange Act'') and the
Investment Advisers Act of 1940 (``Advisers Act'') to eliminate, or
neutralize the effect of, certain conflicts of interest associated with
broker-dealers' or investment advisers' interactions with investors
through these firms' use of technologies that optimize for, predict,
guide, forecast, or direct investment-related behaviors or outcomes.
The Commission is also proposing amendments to rules under the Exchange
Act and Advisers Act that would require firms to make and maintain
certain records in accordance with the proposed conflicts rules.

DATES: Comments should be received on or before October 10, 2023.

ADDRESSES: Comments may be submitted by any of the following methods:

Electronic Comments

<bullet> Use the Commission's internet comment form (<a href="https://www.sec.gov/rules/proposed.shtml">https://www.sec.gov/rules/proposed.shtml</a>); or
<bullet> Send an email to <a href="/cdn-cgi/l/email-protection#493b3c252c642a2624242c273d3a093a2c2a672e263f">[email&#160;protected]</a>. Please include
File Number S7-12-23 on the subject line.

Paper Comments

<bullet> Send paper comments to Vanessa A. Countryman, Secretary,
Securities and Exchange Commission, 100 F Street NE, Washington, DC
20549-1090.

All submissions should refer to File Number S7-12-23. This file number
should be included on the subject line if email is used. To help the
Commission process and review your comments more efficiently, please
use only one method of submission. The Commission will post all
comments on the Commission's website (<a href="https://www.sec.gov/rules/proposed.shtml">https://www.sec.gov/rules/proposed.shtml</a>). Comments are also available for website viewing and
printing in the Commission's Public Reference Room, 100 F Street NE,
Washington, DC 20549, on official business days between the hours of 10
a.m. and 3 p.m. Operating conditions may limit access to the
Commission's Public Reference Room. Do not include personal
identifiable information in submissions; you should submit only
information that you wish to make available publicly. We may redact in
part or withhold entirely from publication submitted material that is
obscene or subject to copyright protection.
Studies, memoranda, or other substantive items may be added by the
Commission or staff to the comment file during this rulemaking. A
notification of the inclusion in the comment file of any such materials
will be made available on the Commission's website. To ensure direct
electronic receipt of such notifications, sign up through the ``Stay
Connected'' option at <a href="http://www.sec.gov">www.sec.gov</a> to receive notifications by email.

FOR FURTHER INFORMATION CONTACT: Blair B. Burnett, Senior Counsel,
Investment Company Regulation Office, Michael Schrader, Senior Counsel,
Chief Counsel's Office, Sirimal R. Mukerjee, Senior Special Counsel,
and Melissa Roverts Harke, Assistant Director, Investment Adviser
Regulation Office, Division of Investment Management, at (202) 551-6787
or <a href="/cdn-cgi/l/email-protection#064f4774736a63754675636528616970">[email&#160;protected]</a>, and Kyra Grundeman and James Wintering, Special
Counsels, Anand Das, Senior Special Counsel, Kelly Shoop, Branch Chief,
Devin Ryan, Assistant Director, John Fahey, Deputy Chief Counsel, and
Emily Westerberg Russell, Chief Counsel, Office of Chief Counsel,
Division of Trading and Markets, at (202) 551-5550 or
<a href="/cdn-cgi/l/email-protection#d1a5a3b0b5b8bfb6b0bfb5bcb0a3bab4a5a291a2b4b2ffb6bea7">[email&#160;protected]</a>, Securities and Exchange Commission, 100 F
Street NE, Washington, DC 20549-8549.

SUPPLEMENTARY INFORMATION: The Commission is proposing for public
comment: 17 CFR 240.15l-2 under the Exchange Act \1\ (``proposed rule
240.151-2'') and 17 CFR 275.211(h)(2)-4 under the Advisers Act \2\
(``proposed rule 275.211(h)(2)-4'' and, together with proposed rule
240.15l-2, ``proposed conflicts rules''); and amendments to 17 CFR
240.17a-3 and 17 CFR 240.17a-4 (``rules 17a-3 and 17a-4'') under the
Exchange Act and 17 CFR 275.204-2 under the Advisers Act (``rule 204-
2'' and, together with the proposed amendments to rules 17a-3 and 17a-
4, ``proposed recordkeeping amendments'').
---------------------------------------------------------------------------

\1\ Unless otherwise noted, when we refer to the Exchange Act,
we are referring to 15 U.S.C. 78, and when we refer to rules under
the Exchange Act, we are referring to title 17, part 240 of the Code
of Federal Regulations [17 CFR 240].
\2\ Unless otherwise noted, when we refer to the Advisers Act,
we are referring to 15 U.S.C. 80b, and when we refer to rules under
the Advisers Act, we are referring to title 17, part 275 of the Code
of Federal Regulations [17 CFR 275].
---------------------------------------------------------------------------

Table of Contents

I. Introduction
A. Overview
B. Background
1. Evolution in the Investment Industry and its Technology Use
2. Current PDA-Like Technology Use and Expected Growth
3. Commission Protection of Investors as Technology Has Evolved
4. Use of Predictive Data Technologies in Investor Interactions
5. Request for Information and Comment
C. Overview of the Proposal
II. Discussion
A. Proposed Conflicts Rules
1. Scope
2. Identification, Determination, and Elimination, or
Neutralization of the Effect of, a Conflict of Interest
3. Policies and Procedures Requirement
B. Proposed Recordkeeping Amendments
III. Economic Analysis
A. Introduction
B. Broad Economic Considerations
C. Economic Baseline
1. Affected Parties
2. Technology and Market Practices
3. Regulatory Baseline
D. Benefits and Costs
1. Benefits
2. Costs
E. Effects on Efficiency, Competition, and Capital Formation
1. Efficiency
2. Competition
3. Capital Formation
F. Reasonable Alternatives
1. Expressly Permit, or Require, the Use of Independent Third-
Party Analyses
2. Require That Senior Firm Personnel and/or Specific Technology
Subject-Matter Experts Participate in the Process of Adopting and
Implementing These Policies and Procedures
3. Provide an Exclusion for Technologies That Consider Large
Datasets Where Firms Have No Reason To Believe the Dataset Favors
the Interests of the Firm From the Identification, Evaluation, and
Testing Requirements
4. Apply the Requirements of the Proposed Conflicts Rule and
Proposed Recordkeeping Amendments Only to Broker-Dealer Use of
Covered Technologies That Have Non-Recommendation Investor
Interaction
5. Require That Firms Test Covered Technologies on an Annual
Basis, or at a Specific Minimum Frequency
6. Require That Firms Provide a Prescribed and Standardized
Disclosure
G. Request for Comment
IV. Paperwork Reduction Act
A. Introduction
B. Proposed Conflicts Rules and Proposed Recordkeeping
Amendments
C. Request for Comment
V. Initial Regulatory Flexibility Analysis

[[Page 53961]]

A. Reason for and Objectives of the Proposed Action
1. Proposed Rules 151-2 and 211(h)(2)-4
2. Proposed Amendments to Rules 17a-3 and 17a-4 and Rule 204-2
B. Legal Basis
C. Small Entities Subject to the Rules and Rule Amendments
1. Small Advisers Subject to Proposed Rule 211(h)(2)-4 and
Proposed Amendments to Recordkeeping Rule
D. Small Broker-Dealers Subject to Proposed Conflicts Rule and
Amendments to Recordkeeping Rules
E. Projected Reporting, Recordkeeping, and Other Compliance
Requirements
1. Proposed Conflicts Rules
2. Proposed Amendments to Rule 204-2
3. Proposed Amendments to Rules 17a-3 and 17a-4
F. Duplicative, Overlapping, or Conflicting Federal Rules
1. Proposed Rule 211(h)(2)-4 and Proposed Amendments to Rule
204-2
2. Proposed Rule 15l-2 and Proposed Amendments to Rules 17a-3
and 17a-4
G. Significant Alternatives
H. Solicitation of Comments
VI. Consideration of Impact on the Economy
Statutory Authority
Text of Proposed Rules and Form Amendments

I. Introduction

The adoption and use of newer technologies, such as predictive data
analytics (``PDA''), by broker-dealers and investment advisers
(together, ``firms'') have accelerated.\3\ In some instances, firms'
use of PDA and similar technologies may be subject to statutory or
regulatory investor protections, but in other cases, it may not. Firms'
use of PDA-like technologies can bring benefits in market access,
efficiency, and returns. To the extent that firms are using PDA-like
technologies to optimize for their own interests in a manner
(intentionally or unintentionally) that places these interests ahead of
investor interests, however, investors can suffer harm. Further, due to
the scalability of these technologies and the potential for firms to
reach a broad audience at a rapid speed, as discussed below, any
resulting conflicts of interest could cause harm to investors in a more
pronounced fashion and on a broader scale than previously possible.\4\
---------------------------------------------------------------------------

\3\ See Deloitte, Artificial intelligence: The next frontier for
investment management firms (Feb. 5, 2019), <a href="https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html">https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html</a> (``AI is
providing new opportunities which extend far beyond cost reduction
and efficient operations. Many investment management firms have
taken note and are actively testing the waters, applying cognitive
technologies and AI to various business functions across the
industry value chain.''); Blake Schmidt and Amanda Albright, AI Is
Coming for Wealth Management. Here's What That Means, Bloomberg
Markets (Apr. 21, 2023), <a href="https://www.bloomberg.com/news/articles/2023-04-21/vanguard-fidelity-experts-explain-how-ai-is-changing-wealth-management">https://www.bloomberg.com/news/articles/2023-04-21/vanguard-fidelity-experts-explain-how-ai-is-changing-wealth-management</a> (discussing experts views on AI impact on the
wealth management industry). As discussed more below, in addition to
PDA, firms have adopted and used artificial intelligence (``AI''),
including machine learning, deep learning, neural networks, natural
language processing (``NLP''), or large language models (including
generative pre-trained transformers or ``GPT''), as well as other
technologies that make use of historical or real-time data, lookup
tables, or correlation matrices (collectively, ``PDA-like
technologies''). See, e.g., Q. Zhu and J. Luo, Generative Pre-
Trained Transformer for Design Concept Generation: An Exploration,
Proceedings of the Design Society, Design Vol 2 (May 2022), <a href="https://www.cambridge.org/core/journals/proceedings-of-the-design-society/article/generative-pretrained-transformer-for-design-concept-generation-an-exploration/41894D82DCBC0610B5B6E68967B7047F">https://www.cambridge.org/core/journals/proceedings-of-the-design-society/article/generative-pretrained-transformer-for-design-concept-generation-an-exploration/41894D82DCBC0610B5B6E68967B7047F</a> (``GPT
are language models pre-trained on vast quantities of textual data
and can perform a wide range of language-related tasks.'')
(citations omitted).
\4\ See infra section I.C.
---------------------------------------------------------------------------

We believe the current regulatory framework should be updated to
help ensure that firms are appropriately addressing conflicts of
interests associated with the use of PDA-like technologies. As a
result, we are proposing specific protections to complement those
already required under existing regulatory frameworks \5\ to better
protect investors from harms arising from these conflicts.
---------------------------------------------------------------------------

\5\ See infra section III.C.3.
---------------------------------------------------------------------------

A. Overview

Broker-dealers may have a range of conflicts of interest with their
retail investors.\6\ Likewise, investment advisers may have conflicts
of interest with respect to advisory clients and investors in their
pooled investment vehicle clients.\7\ Some of these conflicts of
interest are inherent to the relationship between these firms and
investors. For example, an investment adviser that is paid a percentage
fee based on assets under management has an incentive to encourage a
client to move assets into his or her advisory account, which could
conflict with investors' interest, for example, to retain assets in a
401(k) plan or other retirement account. Similarly, a broker-dealer
that receives transaction-based (e.g., commission) compensation has an
incentive to maximize the frequency of transactions, which could
increase costs to the investor or expose them to other risks associated
with excess trading.
---------------------------------------------------------------------------

\6\ While the proposed conflicts rules do not use or define the
term ``retail investors,'' we use that term in this release to mean
``a natural person, or the legal representative of such natural
person, who seeks to receive or receives services primarily for
personal, family or household purposes,'' which is consistent with
the definition of ``retail investor'' in Form CRS and would include
both current and prospective retail customers. See Form CRS, Sec.
11.E. Separately, we note that, for broker-dealers, the proposed
conflicts rule defines ``investor'' consistent with the definition
of ``retail investor'' in Form CRS.
\7\ Proposed rule 275.211(h)(2)-4 would apply to clients and
prospective clients of advisers as well as investors and prospective
investors in pooled investment vehicles advised by those advisers.
---------------------------------------------------------------------------

Many broker-dealers and investment advisers also have conflicts of
interest associated with other common business practices. For example,
some investment product sponsors offer revenue sharing payments,
creating an incentive for broker-dealers and investment advisers that
accept such payments to favor those investments. Similarly, firms that
offer proprietary products have an incentive to favor those products
over other non-proprietary alternatives. Dual registrant and affiliated
firms that offer both brokerage and advisory accounts have an incentive
to steer investors toward the account type that is most profitable for
the firm, regardless of whether it is in the best interest of the
investor. Unless adequately addressed, these conflicts of interest can
cause broker-dealers and investment advisers to place their interests
ahead of investors' interests.
Broker-dealers and investment advisers operate within regulatory
frameworks that in many cases require them to, as applicable, disclose,
mitigate, or eliminate conflicts.\8\ These regulatory frameworks play a
fundamental role in protecting retail investors of broker-dealers,
clients of investment advisers, and investors in pooled investment
vehicle clients of investment advisers (together, ``investors'') from
the negative effects of firms placing their own interests ahead of
investors' interests. As the markets grow and evolve, however, and
specifically, as firms adopt and utilize newer technologies to interact
with investors, we are evaluating our regulations' effectiveness in
protecting investors from the potentially harmful impact of conflicts
of interest.
---------------------------------------------------------------------------

\8\ See <a href="https://www.sec.gov/rules/final/2019/34-86031.pdf">https://www.sec.gov/rules/final/2019/34-86031.pdf</a>,
Exchange Act Release No. 86031 (June 5, 2019) [84 FR 33318 (July 12,
2019)] (``Reg BI Adopting Release''); Commission Interpretation
Regarding Standard of Conduct for Investment Advisers, Advisers Act
Release No. 5248 (June 5, 2019) [84 FR 33669 (July 12, 2019)], at
section II.C. (``Fiduciary Interpretation'') (describing an
adviser's fiduciary duties to its clients). Additionally, rule
206(4)-8 under the Advisers Act prohibits certain statements,
omissions, and other acts, practices, or courses of business as
fraudulent, deceptive, or manipulative with respect to any investor
or prospective investor in a pooled investment vehicle.
---------------------------------------------------------------------------

Recently, firms' adoption and use of PDA-like technologies \9\ have

[[Page 53962]]

accelerated.\10\ While this adoption and use can bring potential
benefits for firms and investors (e.g., with respect to efficiency of
operations, which can generate cost savings for investors, or enhancing
the efficiency of identifying investment opportunities that match an
investor's preferences, profile, and risk tolerances), they also raise
the potential for conflicts of interest associated with the use of
these technologies to cause harm to investors more broadly than
before.\11\
---------------------------------------------------------------------------

\9\ Artificial intelligence is generally used to mean the
capability of a machine to imitate intelligent human behavior and
machine learning is a subfield of artificial intelligence that gives
computers the ability to learn without explicitly being programmed.
See generally Sara Brown, Machine Learning, Explained, MIT Sloan
School of Management (Apr. 21, 2021), <a href="https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained">https://mitsloan.mit.edu/ideas-made-to-matter/machine-learning-explained</a>. Predictive data
analytics draws inferences from large data sets, relying on
hypothesis-free data mining and inductive reasoning to uncover
patterns to make predictions about future outcomes, and may use
natural language processing, signal processing, topic modeling,
pattern recognition, machine learning, deep learning, neural
networks, and other advanced statistical methods. See Nathan Cortez,
Predictive Analytics Law and Policy: Mapping the Terrain:
Challenging Issues in Specific Private Sector Contexts,
Substantiating Big Data in Health Care, 14 ISJLP 61, 65 (Fall 2017).
See generally Financial Industry Regulatory Authority, Inc.
(``FINRA''), Artificial Intelligence (AI) in the Securities Industry
5 (June 2020) (``FINRA AI Report''), <a href="https://www.finra.org/sites/default/files/2020-06/ai-report-061020.pdf">https://www.finra.org/sites/default/files/2020-06/ai-report-061020.pdf</a>; Financial Stability
Board, Artificial Intelligence and Machine Learning in Financial
Services: Market Developments and Financial Stability Implications
(Nov. 1, 2017) (``FSB AI Report''), <a href="https://www.fsb.org/wp-content/uploads/P011117.pdf">https://www.fsb.org/wp-content/uploads/P011117.pdf</a>; see also Department of the Treasury, et al.,
Request for Information and Comment on Financial Institutions' Use
of Artificial Intelligence, Including Machine Learning (Feb. 2021)
[86 FR 16837, 16839-40 (Mar. 31, 2021)] (``Treasury RFI'').
\10\ See infra section I.B.
\11\ See, e.g., For AI in Asset Management, Tomorrow is Here,
Markets Media (Mar. 28, 2023), <a href="https://www.marketsmedia.com/for-ai-in-asset-management-tomorrow-is-here/">https://www.marketsmedia.com/for-ai-in-asset-management-tomorrow-is-here/</a> (citing possible benefits for
investment managers in generating alpha, improving efficiency,
enhancing product and content distribution, and enhancing risk
management and customer experience); Christine Schmid, AI in Wealth:
from Science Fiction to Science Fact, FinExtra (June 8, 2023),
<a href="https://www.finextra.com/blogposting/24323/ai-in-wealth-from-science-fiction-to-science-fact">https://www.finextra.com/blogposting/24323/ai-in-wealth-from-science-fiction-to-science-fact</a> (citing potential benefits in
personalized portfolio creation, enhanced investor engagement,
democratized personalized investing, and reduced information
overload).
---------------------------------------------------------------------------

While the presence of conflicts of interest between firms and
investors is not new, firms' increasing use of these PDA-like
technologies in investor interactions may expose investors to unique
risks. This includes the risk of conflicts remaining unidentified and
therefore unaddressed or identified and unaddressed. The effects of
such unaddressed conflicts may be pernicious, particularly as this
technology can rapidly transmit or scale conflicted actions across a
firm's investor base.\12\ For example, conflicts of interest can arise
from the data the technology uses (including any investor data) and the
inferences the technology makes (including in analyzing that data,
other data, securities, or other assets). These issues may render a
firm's identification of such conflicts for purposes of the firm's
compliance with applicable Federal securities laws more challenging
without specific efforts both to fully understand the PDA-like
technology it is using \13\ and to oversee conflicts that are created
by or transmitted through its use of such technology.\14\
---------------------------------------------------------------------------

\12\ See, e.g., Sophia Duffy and Steve Parrish, You Say
Fiduciary, I Say Binary: A Review and Recommendation of Robo-
Advisors and the Fiduciary and Best Interest Standards, 17 Hastings
Bus. L.J. 3, at 26 (2021) (stating that the impact of firm conflicts
of robo-advisors ``are arguably more detrimental than personal
conflicts between an advisor and client because the number of
clients impacted by the firm conflict is potentially exponentially
higher.'') (``Robo-Advisors and the Fiduciary and Best Interest
Standards'').
\13\ See, e.g., infra section II.A.2.b and II.A.3 (discussing
the testing and policies and procedures requirements, respectively,
of the proposed conflicts rules, which if implemented in accordance
with the proposal, would necessitate firms' developing an
understanding of the PDA-like technologies they use).
\14\ See, e.g., Sohnke M. Bartram, Jurgen Branke & Mehrshad
Motahari, Artificial Intelligence in Asset Management (2020) (``AI
in Asset Management'') (``Understanding and explaining the
inferences made by most AI models is difficult, if not impossible.
As the complexity of the task or the algorithm grows, opacity can
render human supervision ineffective, thereby becoming an even more
significant problem.'').
---------------------------------------------------------------------------

Moreover, PDA-like technologies may have the capacity to process
data, scale outcomes from analysis of data, and evolve at rapid
rates.\15\ While valuable in many circumstances, these technologies
could rapidly and exponentially scale the transmission of any conflicts
of interest associated with such technologies to investors.\16\ For
example, a firm may use PDA-like technologies to automatically develop
advice and recommendations that are then transmitted to investors
through the firm's chatbot, push notifications on its mobile trading
application (``app''), and robo-advisory platform. If the advice or
recommendation transmitted is tainted by a conflict of interest because
the algorithm drifted \17\ to advising or recommending investments more
profitable to the firm or because the dataset underlying the algorithm
was biased toward investments more profitable to the firm, the
transmission of this conflicted advice and recommendations could spread
rapidly to many investors.
---------------------------------------------------------------------------

\15\ See, e.g., Eray Elicik, Artificial Intelligence vs. Human
Intelligence: Can a game-changing technology play the game? (Apr.
20, 2022), <a href="https://dataconomy.com/2022/04/is-artificial-intelligence-better-than-human-intelligence/">https://dataconomy.com/2022/04/is-artificial-intelligence-better-than-human-intelligence/</a> (``Compared to the
human brain, machine learning (ML) can process more data and do so
at a faster rate.''); David Nield, Google Engineers `Mutate' AI to
Make It Evolve Systems Faster Than We Can Code Them (Apr. 17, 2020),
<a href="https://www.sciencealert.com/coders-mutate-ai-systems-to-make-them-evolve-faster-than-we-can-program-them">https://www.sciencealert.com/coders-mutate-ai-systems-to-make-them-evolve-faster-than-we-can-program-them</a> (``[R]esearchers have tweaked
[a machine learning system] to incorporate concepts of Darwinian
evolution and shown it can build AI programs that continue to
improve upon themselves faster than they would if humans were doing
the coding.'').
\16\ See Robo-Advisors and the Fiduciary and Best Interest
Standards, supra note 12, at 26. See also FINRA AI Report, supra
note 9 (discussing exploration of the use of AI tools by market
participants and noting, among other things, that firms should
ensure sound governance and supervision, including effective means
of overseeing suitability of recommendations, conflicts of interest,
customer risk profiles and portfolio rebalancing) (internal
quotations and citation omitted); Y. Minsky, Communications of the
ACM, OCaml for the Masses (Sept. 27, 2011), <a href="https://dl.acm.org/doi/pdf/10.1145/2018396.2018413">https://dl.acm.org/doi/pdf/10.1145/2018396.2018413</a> (explaining that ``technology carries
risk. There is no faster way for a trading firm to destroy itself
than to deploy a piece of trading software that makes a bad decision
over and over in a tight loop'' and that the author's employer seeks
to control these risks by ``put[ting] a very strong focus on
building software that was easily understood--software that was
readable.'').
\17\ See infra note 157 and accompanying text.
---------------------------------------------------------------------------

Unless adequately addressed, the use of these PDA-like technologies
may create or transmit conflicts of interest that place a firm's
interests ahead of investors' interests. This may arise not only when a
firm is providing investment advice or recommendations, but also in the
firm's sales practices and investor interactions more generally, such
as design elements, features, or communications that nudge or prompt
more immediate and less informed action by the investor.\18\ In light
of these developments and risks, and for the reasons we describe
further below, we are proposing that a firm's use of certain PDA-like
technologies in an investor interaction that places the firm's
interests ahead of the investors' interests involves a conflict of
interest that must be eliminated or its effects neutralized in
accordance with the proposed conflicts rules.
---------------------------------------------------------------------------

\18\ See, e.g., CFA Institute, Ethics and Artificial
Intelligence in Investment Management: A Framework for Professionals
(2022) (stating that professionals should ensure they understand the
sources of any potential conflicts generated by the use of
algorithms and work with developers to ensure that such systems do
not inappropriately incorporate fee considerations in the algorithm
generating the investment advice).
---------------------------------------------------------------------------

B. Background

1. Evolution in the Investment Industry and Its Technology Use
Over the last several decades, firms' use of technology to interact
with investors and provide products and services has evolved
significantly, and with it, the nature and extent of the conflicts of
interest this use can create. When Congress first enacted the

[[Page 53963]]

Exchange Act and the Advisers Act, firms were increasingly deploying
what were then considered advanced technologies, such as punch cards
and telex machines. As technology improved, firms began adopting other
technologies, such as computers, email, spreadsheets, and the internet.
The Commission has previously observed that these and other
technologies have helped to promote transparency, liquidity, and
efficiency in our capital markets.\19\ If responsibly implemented and
overseen by firms, new technologies can aid firms' interactions with
investors, and bring greater access and product choice, potentially at
a lower cost, without compromising investor protection, capital
formation, and fair, orderly, and efficient markets.
---------------------------------------------------------------------------

\19\ See Interpretation on Use of Electronic Media, Investment
Company Act Release No. 24426 (Apr. 28, 2000) [65 FR 25843 (May 4,
2000)], at section I; see also Investment Adviser Marketing,
Investment Advisers Act No. 5653 (Dec. 22, 2020) [86 FR 13024 (Mar.
5, 2021)], at section I (``Investment Adviser Marketing Release'')
(noting that the rules are ``designed to accommodate the continual
evolution and interplay of technology and advice'').
---------------------------------------------------------------------------

Where once investors placed trades with their broker in-person,
they eventually began to place orders over the phone, and then through
a website. Now investors can instantaneously place a trade directly
through an app on a smart phone and, instead of a recommendation
delivered by a human, they may receive push notifications potentially
designed to affect trading behavior. These technological interactions
can be designed to respond to human behavior, for example, sending
increased notifications for certain investment products depending on
where the person scrolling through investment products pauses on her
smartphone. As technology continues to evolve, we believe that firms
are likely to increase their reliance on behavioral science frameworks
in influencing investor behavior.\20\ Investors that previously met in
person with their advisers are now able to access computer-generated
advice that is delivered rapidly in an app to many investors by, for
example, a robo-adviser. Rather than advertising in local newspapers,
making cold calls, or relying on referrals, firms are now digitally
targeting investors.\21\
---------------------------------------------------------------------------

\20\ See, e.g., Robert W. Cook, President and CEO of FINRA,
Statement Before the Financial Services Committee U.S. House of
Representatives (May 6, 2021), <a href="https://www.finra.org/media-center/speeches-testimony/statement-financial-services-committee-us-house-representatives">https://www.finra.org/media-center/speeches-testimony/statement-financial-services-committee-us-house-representatives</a> (addressing the ``recent trends of retail trading
platforms is the use of `game-like' and other features that may
encourage investor behaviors'' and ``the growing prevalence of these
features''); Margaret Franklin, Investment Gamification: Not All
Cons, Some Important Pros, Kiplinger (Feb. 20, 2023), <a href="https://www.kiplinger.com/investing/investment-gamification-pros-and-cons">https://www.kiplinger.com/investing/investment-gamification-pros-and-cons</a>
(discussing the use of behavioral techniques and the rising
influence of social media, and stating that the gamification ``style
of trading, ushered in largely by the next generation of investors,
is likely here to stay.''). See also James Tierney, Investment
Games, 72 Duke L.J. 353, 355 (Nov. 2022) (describing the growth of
retail investing and discussing gamification, including how ``mobile
app developers have innovated in user-interface design to compete
with incumbent brokers [by including features such as] intuitive and
appealing design, as well as digital engagement practices that
encourage interaction with the app and that shape the information
users consider in investing,''); Jill E. Fisch, GameStop and the
Reemergence of the Retail Investor, 102 B.U. L. Rev. 1799, 1802
(Oct. 2022) (discussing gamification and the ``evidence that retail
investment and engagement will both continue and evolve.''); Ernst &
Young, Social investing: behavioral insights for the modern wealth
manager (Apr. 2021), <a href="https://www.ey.com/en_us/wealth-asset-management/social-investing-behavioral-insights-for-the-modern-wealth-manager">https://www.ey.com/en_us/wealth-asset-management/social-investing-behavioral-insights-for-the-modern-wealth-manager</a> (``As firms continue to develop social investing
operating models, they can use behavioral science frameworks to
better understand how their client segments are influenced by
digital design and choice architecture[.]'').
\21\ See, e.g., Disclosure Innovations in Advertising and Other
Communications with the Public, FINRA Regulatory Notice 19-31 (Sept.
19, 2019), <a href="https://www.finra.org/rules-guidance/notices/19-31">https://www.finra.org/rules-guidance/notices/19-31</a>; see
also Leslie K. John, Tami Kim, and Kate Barasz, Ads that Don't
Overstep, Harvard Bus. Rev. (Jan.- Feb. 2018), <a href="https://hbr.org/2018/01/ads-that-dont-overstep">https://hbr.org/2018/01/ads-that-dont-overstep</a>.
---------------------------------------------------------------------------

In recent years, we have observed a rapid expansion in firms'
reliance on technology and technology-based products and services.\22\
The use of technology is now central to how firms provide their
products and services to investors.\23\ Some firms and investors in
financial markets now use new technologies such as AI, machine
learning, NLP, and chatbot technologies to make investment decisions
and communicate between firms and investors.\24\ In addition, existing
technologies for data-analytics and data collection continue to improve
and find new applications.\25\
---------------------------------------------------------------------------

\22\ See generally Marc Andreessen, Why Software Is Eating the
World, Wall St. J. (Aug. 20, 2011), <a href="http://www.wsj.com/articles/SB10001424053111903480904576512250915629460">http://www.wsj.com/articles/SB10001424053111903480904576512250915629460</a> (discussing, among other
things, the transformation of the financial services industry by
software over the last 30 years) (``Why Software is Eating the
World''); Robo-Advisors and the Fiduciary and Best Interest
Standards, supra note 12, at 4 (stating that ``[o]ver the past
decade, robo-advisors, or automated systems for providing financial
advice and services, are becoming more and more popular'' and
discussing estimated growth); Nicole G. Iannarone, Fintech's
Promises and Perils Computer as Confidant: Digital Investment Advice
and the Fiduciary Standard, 93 Chi.-Kent L. Rev. 141, 141 (2018)
(``Automated investment advisers permeate the investment industry.
Digital investment advisers are the fastest growing segment of
financial technology (FinTech) and are disrupting traditional
investment advisory delivery models.'') (citations omitted).
\23\ See, e.g., Investment Adviser Marketing Release, supra note
19, at section I (``The concerns that motivated the Commission to
adopt the advertising and solicitation rules [in 1961 and 1979,
respectively] still exist today, but investment adviser marketing
has evolved with advances in technology. In the decades since the
adoption of both the advertising and solicitation rules, the use of
the internet, mobile applications, and social media has become an
integral part of business communications. Consumers today often rely
on these forms of communication to obtain information, including
reviews and referrals, when considering buying goods and services.
Advisers and third parties also rely on these same types of outlets
to attract and refer potential customers.''); FINRA Investor
Education Foundation, Investors in the United States: The Changing
Landscape (Dec. 2022) <a href="https://www.finrafoundation.org/sites/finrafoundation/files/NFCS-Investor-Report-Changing-Landscape.pdf">https://www.finrafoundation.org/sites/finrafoundation/files/NFCS-Investor-Report-Changing-Landscape.pdf</a>
(discussing, among others, website and mobile app use for placing
trades and use of social media sites for obtaining investment
information).
\24\ Michael Kearns & Yuriy Nevmyvaka Machine Learning for
Market Microstructure and High Frequency Trading, High Frequency
Trading--New Realities for Traders, Markets and Regulators (David
Easley, Marcos Lopez de Prado & Maureen O'Hara editors, Risk Books,
2013); see also Christian Thier & Daniel dos Santos Monteiro, How
Much Artificial Intelligence Do Robo-Advisors Really Use? (Aug. 31,
2022), <a href="https://ssrn.com/abstract=4218181">https://ssrn.com/abstract=4218181</a>; Imani Moise, Bond
Investing Gets the Robo-Adviser Treatment, The Wall Street Journal
(June 7, 2023), <a href="https://www.wsj.com/articles/buying-bonds-is-hard-heres-a-way-to-let-a-robot-do-it-70a4587b">https://www.wsj.com/articles/buying-bonds-is-hard-heres-a-way-to-let-a-robot-do-it-70a4587b</a>.
\25\ Natasha Lekh & Petr P[aacute]tek, What's the Future of Web
Scraping in 2023?, APIFY Blog (Jan. 20, 2023), <a href="https://blog.apify.com/future-of-web-scraping-in-2023/">https://blog.apify.com/future-of-web-scraping-in-2023/</a>; Jon Martindale, Best
Apps to Use GPT-4, Digitaltrends (May 4, 2023), <a href="https://www.digitaltrends.com/computing/best-apps-to-use-gpt-4/">https://www.digitaltrends.com/computing/best-apps-to-use-gpt-4/</a>.
---------------------------------------------------------------------------

2. Current PDA-Like Technology Use and Expected Growth

Financial market participants currently use AI and machine learning
technologies in a variety of ways. For example, algorithmic trading is
a widely used application of machine learning in finance, where
machine-learning models analyze large datasets and identify patterns
and signals to optimize for, predict, guide, forecast, or direct
investment-related behaviors or outcomes.\26\ Moreover, the advent and
growth of services available on certain digital platforms, such as
those offered by online brokerages and robo-advisers, have multiplied
the opportunities for retail investors, in particular, to invest and
trade in securities, and in small amounts through fractional
shares.\27\

[[Page 53964]]

This increased accessibility has been one of the key factors associated
with the increase of retail investor participation in U.S. securities
markets in recent years.\28\ Firms have also expanded their use of
technology to include ``digital engagement practices'' or ``DEPs,''
such as behavioral prompts, differential marketing, game-like features
(commonly referred to as ``gamification''), and other design elements
or features designed to engage retail investors when using a firm's
digital platforms (e.g., website, portal, app) \29\ for services such
as trading, robo-advice, and financial education. Our staff has
observed that firms use technology to more efficiently develop
investment strategies, including by using technology to automate their
services, and to analyze the success of specific features and marketing
practices at influencing retail investor behavior.\30\ Firms may also
seek to lower expenses by replacing customer service personnel with
chatbots that can address common customer questions, and outsourcing
their back office operations to vendors that rely heavily on
technology.\31\
---------------------------------------------------------------------------

\26\ See generally Alessio Azzutti, Wolf-Goerge Ringe, H.
Siegfried Stiehl, Machine Learning, Market Manipulation, and
Collusion on Capital Markets: Why the ``Black Box'' Matters, 43 U.
Pa. J. Int'l L. 1 (2021), <a href="https://scholarship.law.upenn.edu/cgi/viewcontent.cgi?article=2035&context=jil">https://scholarship.law.upenn.edu/cgi/viewcontent.cgi?article=2035&context=jil</a> (``Machine Learning and
Market Manipulation'') (discussing current uses of algorithmic
trading and exploring the risks to market integrity in connection
with the evolving uses of artificial intelligence in algorithmic
trading).
\27\ See, e.g., Nolan Schloneger, A Case for Regulating Gamified
Investing, 56 Ind. L. Rev. 175 (2022) (``Th[e] rise [of investing
applications] is largely attributed to zero commission and
fractional-share trading.''); John Csiszar, How Our Approach to
Investing Has Changed Forever, YAHOO! (Mar. 10, 2021), <a href="https://www.yahoo.com/now/approach-investing-changed-forever-190007929.html">https://www.yahoo.com/now/approach-investing-changed-forever-190007929.html</a>
(``Fractional share trading is just in its infancy but appears well
on its way to changing how consumers approach investing. With
fractional share trading, you can invest any dollar amount into
stock, even if you don't have enough to buy a single share . . . .
Fractional share investing allows nearly anyone to get involved in
the stock market without needing $100,000 or more to buy a properly
diversified portfolio of individual stock names.''). See also Staff
Report on Equity and Options Market Structure Conditions in Early
2021 (Oct. 14, 2021), <a href="https://www.sec.gov/files/staff-report-equity-options-market-struction-conditions-early-2021.pdf">https://www.sec.gov/files/staff-report-equity-options-market-struction-conditions-early-2021.pdf</a> (``Some brokers
have sought to attract new customers by offering the ability to
purchase fractional shares. Fractional shares give investors the
ability to purchase less than 1 share of a stock.''). Any staff
statements represent the views of the staff. They are not a rule,
regulation, or statement of the Commission. Furthermore, the
Commission has neither approved nor disapproved their content. These
staff statements, like all staff statements, have no legal force or
effect: they do not alter or amend applicable law; and they create
no new or additional obligations for any person.
\28\ See, e.g., Maggie Fitzgerald, Retail Investors Continue to
Jump Into the Stock Market After GameStop Mania, CNBC (Mar. 10,
2021), <a href="https://www.cnbc.com/2021/03/10/retail-investor-ranks-in-the-stock-market-continue-to-surge.html">https://www.cnbc.com/2021/03/10/retail-investor-ranks-in-the-stock-market-continue-to-surge.html</a> (providing year-over-year app
download statistics for Robinhood, Webull, Sofi, Coinbase, TD
Ameritrade, Charles Schwab, E-Trade, and Fidelity from 2018-2020,
and monthly figures for January and February of 2021); John
Gittelsohn, Schwab Boosts New Trading Accounts 31% After Fees Go to
Zero, Bloomberg (Nov. 14, 2019), <a href="https://www.bloomberg.com/news/articles/2019-11-14/schwab-boosts-brokerage-accounts-by-31-after-fees-cut-to-zero">https://www.bloomberg.com/news/articles/2019-11-14/schwab-boosts-brokerage-accounts-by-31-after-fees-cut-to-zero</a> (noting that Charles Schwab opened 142,000 new
trading accounts in October, a 31% jump over September's pace).
\29\ Examples of DEPs include the following: social networking
tools; games, streaks and other contests with prizes; points,
badges, and leaderboards; notifications; celebrations for trading;
visual cues; ideas presented at order placement and other curated
lists or features; subscriptions and membership tiers; and chatbots.
\30\ See, e.g., SEC Investor Bulletin: Robo-Advisers (Feb. 23,
2017), <a href="https://www.sec.gov/oiea/investor-alerts-bulletins/ib_robo-advisers">https://www.sec.gov/oiea/investor-alerts-bulletins/ib_robo-advisers</a> (discussing automated digital investment advisory
programs); see also FINRA AI Report, supra note 9 (discussing three
areas where broker-dealers are evaluating or using AI in the
securities industry: communications with customers, investment
processes, and operational functions).
\31\ See, e.g., SS&C Gets Automation Rolling with 180 `Digital
Workers', Ignites (Feb. 9, 2023), <a href="https://www.ignites.com/c/3928224/508304?referrer_module=searchSubFromIG&highlight=SS&C">https://www.ignites.com/c/3928224/508304?referrer_module=searchSubFromIG&highlight=SS&C</a>.
---------------------------------------------------------------------------

The rate at which PDA-like technologies continues to evolve is
increasing \32\ and firms are exploring and deploying AI-based
applications across different functions of their organizations,
including customer facing, investment, and operational activities.\33\
These PDA-like technologies are complex and may include several
categories of machine learning \34\ algorithms, such as deep
learning,\35\ supervised learning,\36\ unsupervised learning,\37\ and
reinforcement learning \38\ processes.\39\ In the past few years, these
PDA-like technologies have made increasing use of natural language
processing and natural language generation.\40\ For example, AI has
revolutionized chatbots by enabling them to understand and respond to
natural language more accurately and learn and improve responses over
time, leading to more personalized interactions with users. Recently, a
new wave of online chatbots has rapidly moved machines using AI into
new territory.\41\ Some of these chatbots have passed what is known as
the ``Turing test'' and have become virtually indistinguishable from
humans in particular situations.\42\ AI use is increasing year over
year and in an array of applications.\43\ For instance, some robo-
advisers use chatbots and NLP technology for their online platforms to
provide investment advice and manage investment portfolios.\44\ These
platforms may use a combination of AI, machine learning, NLP, and
chatbot technologies to provide personalized investment recommendations
to customers based on customer risk tolerance and investment goals.
---------------------------------------------------------------------------

\32\ See, e.g., Robin Feldman and Kara Stein, AI Governance in
the Financial Industry, 27 Stan. J.L. Bus. & Fin. 94, 122 (2022)
(describing AI as ``a technology that is rapidly evolving and
capable of learning.'').
\33\ See, e.g., Merav Ozair, FinanceGPT: The Next Generation of
AI-Powered Robo Advisors and Chatbots (June 27, 2023), <a href="https://www.nasdaq.com/articles/financegpt-the-next-generation-of-ai-powered-robo-advisors-and-chatbots">https://www.nasdaq.com/articles/financegpt-the-next-generation-of-ai-powered-robo-advisors-and-chatbots</a> (describing current uses and
development) (``FinanceGPT'').
\34\ FINRA described ``Machine Learning (ML)'' as ``a field of
computer science that uses algorithms to process large amounts of
data and learn from it. Unlike traditional rules-based programming,
[machine learning] models learn from input data to make predictions
or identify meaningful patterns without being explicitly programmed
to do so. There are different types of [machine-learning] models,
depending on their intended function and structure[.]'' See FINRA AI
Report, supra note 9.
\35\ FINRA described a ``deep learning model'' as a model
``built on an artificial neural network, in which algorithms process
large amounts of unlabeled or unstructured data through multiple
layers of learning in a manner inspired by how neural networks
function in the brain. These models are typically used when the
underlying data is significantly large in volume, obtained from
disparate sources, and may have different formats (e.g., text,
voice, and video).'' See id.
\36\ FINRA described a ``supervised machine learning'' as a
model that ``is trained with labeled input data that correlates to a
specified output. . . . The model is continuously refined to provide
more accurate output as additional training data becomes available.
After the model has learned from the patterns in the training data,
it can then analyze additional data to produce the desired output .
. . .'' See id.
\37\ As described by FINRA, in unsupervised machine learning,
``the input data is not labeled nor is the output specified.
Instead, the models are fed large amounts of raw data and the
algorithms are designed to identify any underlying meaningful
patterns. The algorithms may cluster similar data but do so without
any preconceived notion of the output . . . .'' See id.
\38\ As described by FINRA, in reinforcement learning, ``the
model learns dynamically to achieve the desired output through trial
and error. If the model algorithm performs correctly and achieves
the intended output, it is rewarded. Conversely, if it does not
produce the desired output, it is penalized. Accordingly, the model
learns over time to perform in a way that maximizes the net reward .
. . .'' See id.
\39\ See also FSB AI Report, supra note 9; Treasury RFI, supra
note 9.
\40\ See, e.g., FINRA AI Report, supra note 9.
\41\ See Cade Metz, How Smart Are the Robots Getting?, The New
York Times (Jan. 20, 2023, updated Jan. 25, 2023).
\42\ Id. The Turing test is a subjective test determined by
whether the person interacting with a machine believes that they are
interacting with another person. See id.
\43\ Embracing the Rapid Pace of AI, MIT Technology Review
Insights (May 19, 2021), <a href="https://www.technologyreview.com/2021/05/19/1025016/embracing-the-rapid-pace-of-ai/">https://www.technologyreview.com/2021/05/19/1025016/embracing-the-rapid-pace-of-ai/</a>.
\44\ See, e.g., FinanceGPT, supra note 33 (describing current
uses and development).
---------------------------------------------------------------------------

As a result of a growing desire to perform functions remotely and
through automated means, the COVID-19 pandemic accelerated the adoption
of certain PDA-like technologies.\45\ Many

[[Page 53965]]

expect this momentum to continue, with AI becoming a mainstream
technology across many industries, including the financial sector.\46\
Organizations, including firms in the securities industry,\47\ are
using AI in a multitude of ways, including responding to customer
inquiries, automating back-office processes, quality control,\48\ risk
management, client identification and monitoring, selection of trading
algorithms, and portfolio management.\49\ Others are actively
developing investment advisory services based on PDA-like
technologies.\50\ Further, recent advancements in data collection
techniques have significantly enhanced the scale and scope of data
analytics, and its potential applications. Due to increases in
processing power and data storage capacity, a vast amount of data is
now available for high-speed analysis using these technologies.\51\
Furthermore, the range of data types has also expanded, with consumer
shopping histories, media preferences, and online behavior now among
the many types of data that data analytics can use to synthesize
information, forecast financial outcomes, and predict investor and
customer behavior.\52\ Consequently, these technologies can be applied
in novel and powerful ways which may be subtle, such as using the
layout of an app and choice of data presentation and formatting to
influence trading decisions.\53\ Some trading apps use PDA and AI/
machine learning along with detailed user data to increase user
engagement and trading activity.\54\
---------------------------------------------------------------------------

\45\ See, e.g., Joe McKendrick, AI Adoption Skyrocketed Over the
Last 18 Months, Harvard Bus. Rev. (Sept. 27, 2021), <a href="https://hbr.org/2021/09/ai-adoption-skyrocketed-over-the-last-18-months">https://hbr.org/2021/09/ai-adoption-skyrocketed-over-the-last-18-months</a> (``The
[COVID-19] crisis accelerated the adoption of analytics and AI, and
this momentum will continue into the 2020s, surveys show. Fifty-two
percent of companies accelerated their AI adoption plans because of
the Covid crisis, a study by PwC finds. Just about all, 86%, say
that AI is becoming a `mainstream technology' at their company in
2021. Harris Poll, working with Appen, found that 55% of companies
reported they accelerated their AI strategy in 2020 due to Covid,
and 67% expect to further accelerate their AI strategy in 2021.'');
KPMG, Thriving in an AI World: Unlocking the Value of AI Across
Seven Key Industries (May 2021), at 5, <a href="https://advisory.kpmg.us/articles/2021/thriving-in-an-ai-world.html">https://advisory.kpmg.us/articles/2021/thriving-in-an-ai-world.html</a> (``Thriving in an AI
World''); Blake Schmidt and Amanda Albright, AI Is Coming for Wealth
Management. Here's What That Means, Bloomberg Markets (Apr. 21,
2023), <a href="https://www.bloomberg.com/news/articles/2023-04-21/vanguard-fidelity-experts-explain-how-ai-is-changing-wealth-management">https://www.bloomberg.com/news/articles/2023-04-21/vanguard-fidelity-experts-explain-how-ai-is-changing-wealth-management</a>
(discussing experts views on AI impact on the wealth management
industry).
\46\ Id.
\47\ See IOSCO, The use of artificial intelligence and machine
learning by market intermediaries and asset managers (Sept. 2021),
at 1 (``IOSCO AI/ML Report''), <a href="http://iosco.org/library/pubdocs/pdf/IOSCOPD684.pdf">iosco.org/library/pubdocs/pdf/IOSCOPD684.pdf</a> (``Artificial Intelligence (AI) and Machine Learning
(ML) are increasingly used in financial services, due to a
combination of increased data availability and computing power. The
use of AI and ML by market intermediaries and asset managers may be
altering firms' business models.'').
\48\ See Thriving in an AI World, supra note 45; see also FINRA
AI Report, supra note 9, at 5-10 (noting the use of AI in the
securities industry for communications with customers, investment
processes, and operational functions); FINRA, Deep Learning: The
Future of the Market Manipulation Surveillance Program <a href="https://www.finra.org/media-center/finra-unscripted/deep-learning-market-surveillance">https://www.finra.org/media-center/finra-unscripted/deep-learning-market-surveillance</a> (``FINRA's Market Regulation and Technology teams
recently wrapped up an extensive project to migrate the majority of
FINRA's market manipulation surveillance program to using deep
learning in what is perhaps the largest application of artificial
intelligence in the RegTech space to date.''); Machine Learning and
Market Manipulation, supra note 26; IOSCO AI/ML Report, id.
\49\ IOSCO AI/ML Report, supra note 47.
\50\ See, e.g., Hugh Son, JPMorgan is developing a ChatGPT-like
A.I. service that gives investment advice, CNBC (May 25, 2023),
<a href="https://www.cnbc.com/2023/05/25/jpmorgan-develops-ai-investment-advisor.html">https://www.cnbc.com/2023/05/25/jpmorgan-develops-ai-investment-advisor.html</a> (discussing a trademark application filed by JPMorgan
for a product called IndexGPT that will utilize ``cloud computing
software using artificial intelligence'' for ``analyzing and
selecting securities tailored to customer needs[.]'').
\51\ See, e.g., Dimitris Andriosopoulos et al., Computational
Approaches and Data Analytics in Financial Services: A Literature
Review, 70 J. Operational Rsch. Soc. 1581 (2019), <a href="https://doi.org/10.1080/01605682.2019.1595193">https://doi.org/10.1080/01605682.2019.1595193</a>; James Lawler & Anthony Joseph, Big
Data Analytics Methodology in the Financial Industry, 15 Info. Sys.
Ed. J. 38 (July 2017), <a href="https://isedj.org/2017-15/n4/ISEDJv15n4p38.html">https://isedj.org/2017-15/n4/ISEDJv15n4p38.html</a>.
\52\ Daniel Broby, The Use of Predictive Analytics in Finance, 8
J. Fin & Data Sci. 145 (Nov. 2022), <a href="https://doi.org/10.1016/j.jfds.2022.05.003">https://doi.org/10.1016/j.jfds.2022.05.003</a>; OECD, Artificial Intelligence, Machine Learning
and Big Data in Finance: Opportunities, Challenges, and Implications
for Policy Makers (2021), <a href="https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf">https://www.oecd.org/finance/financial-markets/Artificial-intelligence-machine-learning-big-data-in-finance.pdf</a>.
\53\ See, e.g., Sayan Chaudhury and Chinmay Kulkarni, Design
Patterns of Investing Apps and Their Effects on Investing Behaviors
(2021) (``Chaudhury & Kulkarni''), <a href="http://dl.acm.org/doi/fullHtml/10.1145/3461778.3462008">dl.acm.org/doi/fullHtml/10.1145/3461778.3462008</a> (``investing apps can be considered as technical and
social choice architectures that influence investing behavior'').
\54\ See, e.g., Alex McFarland, 10 ``Best'' AI Stock Trading
Bots, Unite.AI (June 4, 2023), <a href="https://www.unite.ai/stock-trading-bots/">https://www.unite.ai/stock-trading-bots/</a>.
---------------------------------------------------------------------------

Any risks of conflicts of interest associated with AI use will
expand as firms' use of AI grows. These risks will have broad
consequences if AI makes decisions that favor the firms' interests and
then rapidly deploys that information to investors, potentially on a
large scale.\55\ Firms' nascent use of AI may already be exposing
investors to these types of risks as well as others.\56\ We are
concerned that firms will intentionally or unintentionally take their
own interest into account in the data or software underlying the
applicable AI, as well as the applicable PDA-like technologies,
resulting in investor harm. Among other things, a firm may use these
technologies to optimize for the firm's revenue or to generate
behavioral prompts or social engineering to change investor behavior in
a manner that benefits the firm but is to the detriment of the
investor.
---------------------------------------------------------------------------

\55\ See, e.g., Robo-Advisors and the Fiduciary and Best
Interest Standards, supra note 12 (stating that the impact of firm
conflicts of robo-advisors ``are arguably more detrimental than
personal conflicts between an advisor and client because the number
of clients impacted by the firm conflict is potentially
exponentially higher.''). See also AI in Asset Management, supra
note 14 (``AI can make wrong decisions based on incorrect inferences
that have captured spurious or irrelevant patterns in the data. For
example, ANNs [artificial neural networks] that are trained to pick
stocks with high expected returns might select illiquid, distressed
stocks.''); FINRA AI Report, supra note 9, at 11-19 (noting that the
use of AI ``raises several concerns that may be wide-ranging across
various industries as well as some specific to the securities
industry. Over the past few years, there have been numerous
incidents reported about AI applications that may have been
fraudulent, nefarious, discriminatory, or unfair, highlighting the
issue of ethics in AI applications.''); FINRA AI Report, supra note
9, at 13 (``Depending on the use case, data scarcity may limit the
model's analysis and outcomes, and could produce results that may be
narrow and irrelevant. On the other hand, incorporating data from
many different sources may introduce newer risks if the data is not
tested and validated, particularly if new data points fall outside
of the dataset used to train the model.'').
\56\ See, e.g., FINRA AI Report, supra note 9, at 5 (``The use
of AI-based applications is proliferating in the securities
industry[.]''); Sophia Duffy and Steve Parrish, You Say Fiduciary, I
Say Binary: A Review and Recommendation of Robo-Advisors and the
Fiduciary and Best Interest Standards, 17 Hastings Bus. L.J. 3, at
26 (2021) (``robo-advisors can be, and often are, intentionally
programmed to favor the institution by making recommendations that
favor the institution's products, rebalance client portfolios in
ways which will allow the institution to earn more fees, and
otherwise make recommendations that benefit the firm'').
---------------------------------------------------------------------------

3. Commission Protection of Investors as Technology Has Evolved
As noted above, firms' use of technology and subsequent adaptation
incorporating emerging technologies are not new.\57\ At the same time,
the Commission has addressed firms' relationships with investors in a
variety of ways to ensure investor protection as use of technology in
those relationships has evolved over time.\58\ The proposal, thus, is
consistent with the Commission's practice of evolving our regulation in
light of market and technological developments.
---------------------------------------------------------------------------

\57\ See supra section I.B.2.
\58\ See infra note 114.
---------------------------------------------------------------------------

Broker-dealers and investment advisers are currently subject to
extensive obligations under Federal securities laws and regulations,
and, in the case of broker-dealers, rules of self-regulatory
organizations,\59\ that are

[[Page 53966]]

designed to promote conduct that, among other things, protects
investors, including protecting investors from conflicts of
interest.\60\ To the extent PDA-like technologies are used in investor
interactions that are subject to existing obligations, those
obligations apply. These obligations include, but are not limited to,
obligations related to investment advice and recommendations; \61\
general and specific requirements aimed at addressing certain conflicts
of interest, including requirements to eliminate, mitigate, or disclose
certain conflicts of interest; disclosure of firms' services, fees, and
costs; disclosure of certain business practices, advertising,
communications with the public (including the use of ``investment
analysis tools''); supervision; and obligations related to policies and
procedures.\62\ In addition to these obligations, Federal securities
laws and regulations broadly prohibit fraud by broker-dealers and
investment advisers as well as fraud by any person in the offer,
purchase, or sale of securities, or in connection with the purchase or
sale of securities.
---------------------------------------------------------------------------

\59\ Any person operating as a ``broker'' or ``dealer'' in the
U.S. securities markets must register with the Commission, absent an
exception or exemption. See Exchange Act section 15(a), 15 U.S.C.
78o(a); see also Exchange Act sections 3(a)(4) and 3(a)(5), 15
U.S.C. 78c(a)(4) and 78c(a)(5) (definitions of ``broker'' and
``dealer,'' respectively). Generally, all registered broker-dealers
that deal with the public must become members of FINRA, a registered
national securities association, unless the broker or dealer effects
transactions in securities solely on an exchange of which it is a
member. See Exchange Act section 15(b)(8), 15 U.S.C. 78o(b)(8); see
also 17 CFR 240.15b9-1 (providing an exemption from Section
15(b)(8)). FINRA is the sole national securities association
registered with the SEC under Section 15A of the Exchange Act.
Because this release is focused on broker-dealers that deal with the
public and are FINRA member firms (unless an exception applies), we
refer to FINRA rules as broadly applying to ``broker-dealers,''
rather than to ``FINRA member firms.''
\60\ See infra section III.C.3; Fiduciary Interpretation, supra
note 8, at section II.C. (``The duty of loyalty requires that an
adviser not subordinate its clients' interests to its own.''); see
also Reg BI Adopting Release, supra note 8, at section II.A.1. (The
``without placing the financial or other interest . . . ahead of the
interest of the retail customer'' phrasing recognizes that while a
broker-dealer will inevitably have some financial interest in a
recommendation--the nature and magnitude of which will vary--the
broker-dealer's interests cannot be placed ahead of the retail
customer's interest''). Additionally, broker-dealers often provide a
range of services that do not involve a recommendation to a retail
customer--which is required in order for Reg BI to apply--and those
services are subject to general and specific requirements to address
associated conflicts of interest under the Exchange Act, Securities
Act of 1933, and relevant self-regulatory organization (``SRO'')
rules as applicable. See also FINRA Report on Conflicts of Interest
(Oct. 2013), at Appendix I (Conflicts Regulation in the United
States and Selected International Jurisdictions) (``FINRA Conflict
Report''), <a href="https://www.finra.org/sites/default/files/Industry/p359971.pdf">https://www.finra.org/sites/default/files/Industry/p359971.pdf</a> (describing broad obligations under SEC and FINRA rules
as well as specific conflicts-related disclosure requirements under
FINRA rules).
\61\ See, e.g., 17 CFR 240.15l-1(a)(1) (``Exchange Act rule 15l-
1(a)(1)'') (requiring broker-dealers and their associated persons to
act in the best interest of retail customers when making
recommendations, without placing the financial or other interest of
the broker-dealer or its associated person ahead of the interest of
the retail customer).
\62\ Compliance with the proposed conflicts rules would not
alter a broker-dealer's or investment adviser's existing obligations
under the Federal securities laws. The proposed conflicts rules
would apply in addition to any other obligations under the Exchange
Act and Advisers Act, along with any rules the Commission may adopt
thereunder, and any other applicable provisions of the Federal
securities laws and related rules and regulations.
---------------------------------------------------------------------------

The Commission has long acted to protect investors against the harm
that can come when a firm acts on its conflicts of interest.\63\ For
example, the Commission has brought enforcement actions regarding an
investment adviser's fiduciary duty to its clients with respect to
conflicts of interest.\64\ Similarly, the Commission has reinforced
fraud protection for investors in pooled investment vehicles against
conflicts of interest through rule 206(4)-8.\65\ The Commission
regulates investment adviser advertising and marketing practices to
protect against, among others, adviser conflicts of interest that may
taint such marketing, including through recent amendments adapting
those protections in light of the evolution of practices and
technologies.\66\
---------------------------------------------------------------------------

\63\ See infra section III.C.
\64\ See, e.g., SEC Press Release, SEC Share Class Initiative
Returning More Than $125 Million to Investors: Reflecting SEC's
Commitment to Retail Investors, 79 Investment Advisers Who Self-
Reported Advisers Act Violations Agree to Compensate Investors
Promptly, Ensure Adequate Fee Disclosures (Mar. 11, 2019), <a href="https://www.sec.gov/news/press-release/2019-28">https://www.sec.gov/news/press-release/2019-28</a> (describing settled orders
against 79 investment advisers finding that the settling investment
advisers placed their clients in mutual fund share classes that
charged 12b-1 fees when lower-cost share classes of the same fund
were available to their clients without adequately disclosing that
the higher cost share class would be selected; according to the
SEC's orders, the 12b-1 fees were routinely paid to the investment
advisers in their capacity as brokers, to their broker-dealer
affiliates, or to their personnel who were also registered
representatives, creating a conflict of interest with their clients,
as the investment advisers stood to benefit from the clients' paying
higher fees); SEC v. Sergei Polevikov, et al., Litigation Release
No. 25475 (Aug. 17, 2022) (settled order) (final judgment against
employee working as a quantitative analyst at two asset management
firms ``for perpetrating a front-running scheme that generated
profits of approximately $8.5 million''); SEC Brings Settled Actions
Charging Cherry-Picking and Compliance Failures, Adm. Proc. File No.
3-20955 (Aug 10, 2022) (settled order) (alleged multi-year cherry-
picking scheme of former investment adviser representative of
registered investment adviser preferentially allocating profitable
trades or failing to allocate unprofitable trades to a adviser's
personal accounts at the expense of the advisers client accounts).
\65\ 17 CFR 275.206(4)-8; see, e.g., In re. Virtua Capital
Management, LLC, et al., Advisers Act Release No. 6033 (May 23,
2022) (allegedly failing to disclose conflicts of interest and
associated fees, and breaching fiduciary duty to multiple private
investment funds) (settled order).
\66\ See Investment Adviser Marketing Release, supra note 19, at
section I (``The concerns that motivated the Commission to adopt the
advertising and solicitation rules [in 1961 and 1979, respectively]
still exist today, but investment adviser marketing has evolved with
advances in technology. In the decades since the adoption of both
the advertising and solicitation rules, the use of the internet,
mobile applications, and social media has become an integral part of
business communications. Consumers today often rely on these forms
of communication to obtain information, including reviews and
referrals, when considering buying goods and services. Advisers and
third parties also rely on these same types of outlets to attract
and refer potential customers.'').
---------------------------------------------------------------------------

Likewise, broker-dealers have long been subject to Commission and
SRO regulations and rules that govern their business conduct, including
general and specific obligations to address conflicts of interest.\67\
For example, under existing antifraud provisions of the Exchange Act, a
broker-dealer has a duty to disclose material adverse information to
its customers.\68\ Indeed, the Commission has enforced a broker-
dealer's duty to disclose material conflicts of interest under the
antifraud provisions.\69\ Broker-dealers are subject to specific FINRA
rules aimed at addressing certain conflicts of interest.\70\ Moreover,
in 2019 the Commission adopted Regulation Best Interest (``Reg BI''),
which was designed to enhance the quality of broker-dealer
recommendations to retail customers and reduce the potential harm to
retail customers that may be caused by conflicts of interest,\71\ by
requiring broker-dealers that make recommendations to retail customers
to, among other things, establish, maintain, and enforce policies and
procedures reasonably designed to identify and

[[Page 53967]]

disclose, mitigate, or eliminate, conflicts associated with a
recommendation, including conflicts of interest that may result through
the use of PDA-like technology to make recommendations (Reg BI's
``Conflict of Interest Obligation'').\72\
---------------------------------------------------------------------------

\67\ See infra section III.C.3
\68\ A broker-dealer may be liable if it does not disclose
``material adverse facts of which it is aware.'' See, e.g., Chasins
v. Smith, Barney & Co., 438 F.2d 1167, 1172 (2nd Cir. 1970); SEC v.
Hasho, 784 F. Supp. 1059, 1110 (S.D.N.Y. 1992); In the Matter of
RichMark Capital Corp., Exchange Act Release No. 48758 (Nov. 7,
2003) (Commission Opinion) (``When a securities dealer recommends
stock to a customer, it is not only obligated to avoid affirmative
misstatements, but also must disclose material adverse facts of
which it is aware. That includes disclosure of `adverse interests'
such as `economic self-interest' that could have influenced its
recommendation.'') (citations omitted).
\69\ See, e.g., In re. Edward D. Jones & Co, Securities Act
Release No. 8520 (Dec. 22, 2004) (settled order) (broker-dealer
violated antifraud provisions of Securities Act and Exchange Act by
failing to disclose conflicts of interest arising from receipt of
revenue sharing, directed brokerage payments and other payments from
``preferred'' families that were exclusively promoted by broker-
dealer); In re. Morgan Stanley DW Inc., Securities Act Release No.
8339 (Nov. 17, 2003) (settled order) (broker-dealer violated
antifraud provisions of Securities Act by failing to disclose
special promotion of funds from families that paid revenue sharing
and portfolio brokerage).
\70\ FINRA rules establish restrictions on the use of non-cash
compensation in connection with the sale and distribution of mutual
funds, variable annuities, direct participation program securities,
public offerings of debt and equity securities, investment company
securities, real estate investment trust programs, and the use of
non-cash compensation to influence or reward employees of others.
See FINRA Rules 2310, 2320, 2331, 2341, 5110, and 3220. These rules
generally limit the manner in which members can pay or accept non-
cash compensation and detail the types of non-cash compensation that
are permissible.
\71\ See Reg BI Adopting Release supra note 8, at text
accompanying n.21.
\72\ 17 CFR 240.15l-1(a)(2)(iii) (``Exchange Act rule 15l-
1(a)(2)(iii)'').
---------------------------------------------------------------------------

The Commission has and will continue to bring enforcement actions
for violations of the Federal securities laws that entail the use of
PDA-like technologies. However, the rapid acceleration of PDA-like
technologies and their adoption in the investment industry,\73\ the
additional challenges associated with identifying and addressing
conflicts of interest resulting from the use of these new technologies,
and the concerns relating to scalability, discussed above, reinforce
the importance of ensuring our regulatory regime specifically addresses
these issues. In particular, disclosure may be ineffective in light of,
as discussed above, the rate of investor interactions, the size of the
datasets, the complexity of the algorithms on which the PDA-like
technology is based, and the ability of the technology to learn
investor preferences or behavior, which could entail providing
disclosure that is lengthy, highly technical, and variable, which could
cause investors difficulty in understanding the disclosure.
---------------------------------------------------------------------------

\73\ See, e.g., Amy Caiazza, Rob Rosenblum, and Danielle
Sartain, Investment Advisers' Fiduciary Duties: The Use of
Artificial Intelligence, Harvard Law School Forum on Corporate
Governance (June 11, 2020), <a href="https://corpgov.law.harvard.edu/2020/06/11/investment-advisers-fiduciary-duties-the-use-of-artificial-intelligence/">https://corpgov.law.harvard.edu/2020/06/11/investment-advisers-fiduciary-duties-the-use-of-artificial-intelligence/</a> (``Artificial intelligence (AI) is an increasingly
important technology within the investment management industry.'');
FINRA AI Report, supra note 9, at 5 (``The use of AI-based
applications is proliferating in the securities industry and
transforming various functions within broker-dealers.'').
---------------------------------------------------------------------------

In light of these concerns, and the harm to investors that can
result when firms act on conflicts of interest, we are proposing rules
to address conflicts of interest associated with a firm's use of PDA-
like technologies when interacting with investors that are contrary to
the public interest and the protection of investors. In particular, the
recent and rapid expansion of PDA-like technologies in the context of
investment-related activities, without specific oversight obligations
tailored to the specific risks involved in their use, can lead to
outcomes that financially benefit firms at the expense of investors.
Such a harm to investors might include the use of PDA-like technologies
that prompt investors to enroll in products or services that
financially benefit the firm but may not be consistent with their
investment goals or risk tolerance, encourage investors to enter into
more frequent trades or employ riskier trading strategies (e.g., margin
trading) that will increase the firm's profit at the investors'
expense, or inappropriately steer investors toward complex and risky
securities products inconsistent with investors' investment objectives
or risk profiles that result in harm to investors but that financially
benefit the firm. Due to the inherent complexity and opacity of these
technologies as well as their potential for scaling, we are proposing
that such conflicts of interest should be eliminated or their effects
should be neutralized, rather than handled by other methods of
addressing the conflicts, such as through disclosure and consent.
Moreover, many of these technologies provide means--for example, A/B
testing \74\--to empirically assess the conflicts' impact and thus to
neutralize the effect of a conflict on investors. Further, reliance on
scalable, complex, and opaque PDA-like technologies can result in
operational challenges or shortcomings. For example, failure to
identify and address conflicts that may be present in the PDA-like
technology used to steer investors toward a product or service could
result in a firm's failure to identify the risks to investors of
certain investing behaviors that place the firm's interest ahead of
investors' interest as well as inadequate compliance policies and
procedures that would assist the firm in curbing these practices. As a
consequence, this could result in the failure to take sufficient steps
to address the potentially harmful effect of those conflicts.\75\ For
these additional reasons, we are proposing that such conflicts of
interest be eliminated or their effects be neutralized, rather than
handled by other methods of addressing the conflicts, such as through
disclosure and consent.
---------------------------------------------------------------------------

\74\ A/B testing refers to running a learning model on two
different datasets with a single change between the two, which can
help identify causal relationships and, through understanding how
changes affect outcomes, gain a better understanding of the
functionality of a model. See Seldon, A/B Testing for Machine
Learning (July 7, 2021) (``Seldon''), <a href="https://www.seldon.io/a-b-testing-for-machine-learning">https://www.seldon.io/a-b-testing-for-machine-learning</a>.
\75\ See, e.g., William Shaw and Aisha S. Gani, Wall Street
Banks Seizing AI to Rewire the World of Finance, Financial Review
(June 1, 2023) (in discussing fiduciary duty obligation when using
AI in finance quoting a law firm partner as saying: ``How do you
demonstrate to investors and regulators that you've done your duty
when you've used an output without really knowing what the inputs
are?'').
---------------------------------------------------------------------------

4. Use of Predictive Data Technologies in Investor Interactions
Firms may use PDA-like technologies to transform user interfaces
and the interactions that investors have on digital platforms.\76\ For
example, firms may collect data from a variety of internal sources
(e.g., trading desks, customer account histories, and communications)
and external sources (e.g., public filings, social media platforms, and
satellite images) in both structured and unstructured formats,\77\
enabling them to develop an understanding of investor preferences and
adapt the interface and related prompts to appeal to those preferences.
Firms may use these tools to increase the quantity of information used
to support investment ideas,\78\ leverage investor data to send
targeted questionnaires to investors regarding evolving investment
goals, identify which investors might be open to a new investment
product, or identify which investors are most likely to stop using a
firm's services.\79\ We are concerned, however, that a firm's use of
PDA-like technologies when engaging or communicating with--including by
providing information to, providing recommendations or advice to, or
soliciting--a prospective or current investor could take into
consideration the firm's interest in a manner that places its interests
ahead of investors' interests and thus harm investors.\80\ For example,
some members of the public have expressed concern that firms' use of
these PDA-like technologies encourages practices that are profitable
for the firm but may increase investors' costs, undermine investors'
performance, or expose investors to

[[Page 53968]]

unnecessary risks based on their individual investment profile, such
as: (i) excessive trading,\81\ (ii) using trading strategies that carry
additional risk (e.g., options trading and trading on margin), and
(iii) trading in complex securities products that are more remunerative
to the firm but pose undue risk to the investor.\82\
---------------------------------------------------------------------------

\76\ See, e.g., FSB AI Report, supra note 9, at 14-15 (chatbots
are being introduced by a range of financial services firms, often
in mobile apps or social media, and chatbots are ``increasingly
moving toward giving advice and prompting customers to act'').
\77\ See FINRA AI Report, supra note 9, at 4.
\78\ See Deloitte, Artificial intelligence: The next frontier
for investment management firms (Feb. 5, 2019), <a href="https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html">https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html</a>.
\79\ See Ryan W. Neal, Three Firms Where Artificial Intelligence
is Helping with Financial Planning (Jan. 17, 2020), <a href="https://www.investmentnews.com/artificial-intelligence-advisers-176541">https://www.investmentnews.com/artificial-intelligence-advisers-176541</a>
(describing current uses of AI and their potential application to
broker-dealers and investment advisers).
\80\ While the proposed rules apply more broadly to the use of
covered technology in investor interactions, as discussed below,
firms using covered technology to provide advice or make
recommendations are subject to standards of conduct, among other
regulatory obligations, that already apply to such advice or
recommendations. See infra section III.C.3. The proposed conflicts
rules would apply in addition to these standards of conduct and
other regulatory obligations.
\81\ See, e.g., Comment Letter from Pace Investor Rights Clinic
(Oct. 1, 2021) (``Pace University Letter'') (``DEPs can lead
investors to trade more frequently and more often than is in their
best interest. For example, the push notification feature provides
investors with live price updates. This intentionally prompts
investors to check their portfolios after receiving the
notification, which can lead them to make additional trades or spend
more time on the platform than they would have otherwise.
Traditionally, the goal of investing for most retail investors is to
save for the long term. Frequently checking their portfolio may
cause investors to make decisions not in line with the goal of long-
term saving and generational wealth building.''). See also, e.g.,
Feedback Flyer Response of Lincoln Li on S7-10-21 (Aug. 27, 2021)
(``I started half a decade ago following value investing practices.
However, [online investment and trading apps], that I used for a
short time got me into day trading and speculation more frequently.
I ended up stopping using these apps because they took up so much
time with little gain. I spent more time long term trading based off
of proper market factors and evaluation. There's a big concern to
me, especially as a professional game designer, as to how
gamification in life impacting subjects can have negative impact on
society, culture and personal finances. I have friends who got into
technical trading and day trading due to these apps, who talk more
like gamblers than actual investors. It sets a very poor precedent
for this industry and behavior.''); Feedback Flyer Response of
Richard Green on S7-10-21 (Sept. 25, 2021) (responding to a question
about online trading and investment platforms: ``[m]y broker rewards
referrals by offering free stocks for each referral. I think this
pulls new investors into trading, which makes a lot of money for the
broker, as newer investors are more likely to trade too frequently
or make mistakes.''); Feedback Flyer Response of Joseph on S7-10-21
(Aug. 28, 2021) (``[A trading app's] user interface is set up in a
way to subconsciously influence retail traders to trade more
frequently and engage in riskier investment products (options) than
the average amount.'').
\82\ In Congressional hearings related to market events in
January 2021, investor protection concerns were identified relating
to the use of certain types of DEPs, including advertisements
targeted towards specific groups of investors on digital platforms
and game-like features on mobile apps. See Game Stopped? Who Wins
and Loses When Short Sellers, Social Media, and Retail Investors
Collide: Hearing Before the H. Comm. on Fin. Servs., 113th Cong.
(2021), <a href="https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=407107">https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=407107</a>; Game Stopped? Who Wins and Loses
When Short Sellers, Social Media, and Retail Investors Collide, Part
II: Hearing Before the H. Comm. on Fin. Servs., 113th Cong. (2021),
<a href="https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=406268">https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=406268</a>, Game Stopped? Who Wins and Loses
When Short Sellers, Social Media, and Retail Investors Collide, Part
III: Hearing Before the H. Comm. on Fin. Servs., 113th Cong. (2021),
<a href="https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=407748">https://financialservices.house.gov/calendar/eventsingle.aspx?EventID=407748</a>; Who Wins on Wall Street? GameStop,
Robinhood, and the State of Retail Investing: Hearing Before the S.
Comm. On Banking, Hous., & Urban Affairs, 113th Cong. (2021),
<a href="https://www.banking.senate.gov/hearings/who-wins-on-wall-street-gamestop-robinhoodand-the-state-of-retail-investing">https://www.banking.senate.gov/hearings/who-wins-on-wall-street-gamestop-robinhoodand-the-state-of-retail-investing</a>.
---------------------------------------------------------------------------

In some cases, the use of PDA-like technologies to place a firm's
interests ahead of investors' interests could reflect an intentional
design choice.\83\ In other cases, however, the actions that place a
firm's interests ahead of the interest of investors may instead reflect
the firm's failure to fully understand the effects of its use of PDA-
like technologies or to provide appropriate oversight of its use of
such technologies.\84\ For example, AI and other similar technology are
only as good as the data upon which it is based. Corrupted or
mislabeled data, biased data, or data from unknown sources, can
undermine data quality, leading to skewed outcomes with opaque biases
as well as unintended failures.\85\
---------------------------------------------------------------------------

\83\ See, e.g., Megan Ji, Note, Are Robots Good Fiduciaries?
Regulating Robo-Advisors Under the Investment Advisers Act of 1940,
117 Colum. L. Rev. 1543, 1580 (Oct. 2017) (recommending that the
Commission adopt regulations in which ``robo-advisors, in their
disclosures, clearly delineate between conflicts that are programmed
into their algorithms and conflicts that may affect the design of
algorithms.'').
\84\ See Catherine Thorbecke, Plagued with errors: A news
outlet's decision to write stories with AI backfires, CNN (Jan. 23,
2023), <a href="https://www.cnn.com/2023/01/25/tech/cnet-ai-tool-news-stories/index.html">https://www.cnn.com/2023/01/25/tech/cnet-ai-tool-news-stories/index.html</a>.
\85\ See, e.g., Regulation Systems Compliance and Integrity,
Release No. 34-97143 (Mar. 15, 2023) [88 FR 23146 (Apr. 14, 2023)]
(describing the potential market impact of a corrupted data
security-based swap data repository). See also National Institute of
Science and Technology Special Publication 1270, Towards a Standard
for Identifying and Managing Bias in Artificial Intelligence (Mar.
2022), at section 3.1 (describing dataset challenges resulting in AI
bias, discrimination, and systematic gaps in performance); Thor
Olavsrud, 7 famous analytics and AI disasters (Apr. 15, 2022),
<a href="https://www.cio.com/article/190888/5-famous-analytics-and-ai-disasters.html">https://www.cio.com/article/190888/5-famous-analytics-and-ai-disasters.html</a>.
---------------------------------------------------------------------------

While the risk of poor data quality or skewed data is not unique to
AI, the ability of PDA-like technologies used in investor interactions
to process data more quickly than humans, and the potential for
technology to disseminate the resulting communications to a mass
market, can quickly magnify conflicts of interest and any resulting
negative effects on investors. Moreover, erroneous data considered by a
firm's algorithm could have the effect of optimizing for the firm's
interest over investors' interest by, for example, relying on outdated,
previously higher cost information of investment options sponsored by
other firms but relying on updated, lower cost information of identical
investment options sponsored by the firm. This could result in a
recommendation, advice, or other investor interaction that favors the
firm's sponsored products and creates a conflict, regardless of whether
the firm intentionally developed the algorithm to optimize for its
interest.\86\ Poor data quality or skewed data could not only limit the
learning capability of an AI or machine learning system but could also
potentially negatively impact how it makes inferences and decisions in
the future,\87\ giving rise to erroneous or poor predictions, resulting
in a failure to achieve the system's intended objectives,\88\ and
benefiting the firm over investors (whether intentionally or
unintentionally).
---------------------------------------------------------------------------

\86\ In this example, it is also possible that erroneous data
could result in the reverse effect, generating a recommendation in
favor of a non-sponsored product when the firm's sponsored product
may be more cost-effective. This would not result in a conflict
under the proposed rules but would nonetheless be subject to firms'
obligations under their respective regulatory regimes, including the
applicable standard of conduct.
\87\ See Artificial Intelligence/Machine Learning Risk &
Security Working Group (AIRS), Artificial Intelligence Risk &
Governance, at 2.1.1 (accessed Apr. 18, 2023) (``AIRS White
Paper''), <a href="https://aiab.wharton.upenn.edu/research/artificial-intelligence-risk-governance/">https://aiab.wharton.upenn.edu/research/artificial-intelligence-risk-governance/</a>.
\88\ Id.
---------------------------------------------------------------------------

We have observed instances where conflicts of interest associated
with a firm's use of PDA-like technologies have resulted in harm to
investors. A recent enforcement action involved allegations that an
adviser marketed that its ``no fee'' robo-adviser portfolios were
determined through a ``disciplined portfolio construction methodology''
when they allegedly were pre-set to hold a certain percent of assets in
cash because the adviser's affiliate was guaranteed a certain amount of
revenue at these levels. The adviser allegedly did not disclose its
conflict of interest in setting the cash allocations; that this
conflict resulted in higher cash allocations, which could negatively
impact performance in a rising market; and that the cash allocations
were higher than other services because clients did not pay a fee.\89\
While the focus of that action was on the alleged disclosure failure,
it also highlights the potential for PDA-like technologies to be used
in ways that advance a firm's interests at the expense of its
investors' interests. The proposed conflicts rules would require a firm
to analyze its investor interactions that use PDA-like technology for
the types of conflicts of interest that were at issue in that action in
order to determine whether the investor interaction places the firm's
interests ahead of its investors' interests and, if so, eliminate, or
neutralize the effect of, the conflicts of interest on investors. In
addition, the Commission's 2021 Request for Information and Comments on
Broker-Dealer and

[[Page 53969]]

Investment Adviser Digital Engagement Practices, Related Tools and
Methods, and Regulatory Considerations and Potential Approaches
(``Request'') \90\ solicited comments related to conflicts of interest,
among other areas.\91\ In response, the Commission received comments
reflecting perceived conflicts of interest related to the use of online
investing and trading applications, which some commenters indicated
undermine their faith in the fairness of the markets.\92\
---------------------------------------------------------------------------

\89\ In re. Charles Schwab & Co., Inc., et al., Exchange Act
Release No. 95087 (June 13, 2022) (settled order).
\90\ See Request for Information and Comments on Broker-Dealer
and Investment Adviser Digital Engagement Practices, Related Tools
and Methods, and Regulatory Considerations and Potential Approaches,
Exchange Act Release No. 92766 (Aug. 27, 2021) [86 FR 49067 (Sept.
1, 2021)].
\91\ See id., questions 1.26, 2.6, 3.5, 3.16, and 4.15. For
additional discussion regarding the Request, see infra section I.B.5
\92\ See, e.g., Feedback Flyer Response of Tomas Liutvinas on
S7-10-21 (Aug. 28, 2021) (``It seems like there is no conflict of
interest regulations in the US financial system. This makes me
uneasy. Until the rights are fully explained, reported, and undone I
will recommend to anyone I know to stay away from US markets. For
myself, I've invested in a certain position with plans to leave the
investment for the future generations of my family, to hold on
hopefully up to a point when markets will be made transparent and
fair.''); Feedback Flyer Response of Jasper Pummell on S7-10-21
(Aug. 28, 2021) (``I believe that online brokerages have a conflict
of interest and financial regulation is needed to ensure that the
markets are a safe place for retail traders.''); Feedback Flyer
Response of Robert on S7-10-21 (Aug. 27, 2021) (``Retail needs a
fair and transparent market. There are blantant [sic] conflicts of
interest in the market which should be rectified immediately.
Failure to do so will have a mass exodus of investors from the US
stock market.''). See also FINRA AI Report, supra note 9, at 11
(``However, use of AI also raises several concerns that may be wide-
ranging across various industries as well as some specific to the
securities industry. Over the past few years, there have been
numerous incidents reported about AI applications that may have been
fraudulent, nefarious, discriminatory, or unfair, highlighting the
issue of ethics in AI applications.''). But see, e.g., Comment
Letter from David Dusseault, President, Robinhood Financial, LLC
(Oct. 1, 2021) (``Robinhood Letter'') (stating that conflicts of
interest are not new to the financial industry and that the
regulatory frameworks established by the SEC, such as Reg BI and the
disclosure requirements of the Investment Advisers Act of 1940, rest
on the principle that conflicts of interest exist, but investors are
able to navigate them when they are adequately disclosed); Comment
Letter from Investment Adviser Association (Oct. 1, 2021) (``IAA
Letter''); Comment Letter from Kevin M. Carroll, Managing Director
and Associate General Counsel, Securities Industry and Financial
Markets Association (Oct. 1, 2021) (``SIFMA Letter'') (generally
opposing new rules, guidance, or interpretations to address the use
of digital engagement practices). These comments are all available
in the comment file at <a href="https://www.sec.gov/comments/s7-10-21/s71021.htm">https://www.sec.gov/comments/s7-10-21/s71021.htm</a>.
---------------------------------------------------------------------------

Failures to appropriately oversee these PDA-like technologies
compound the risk that conflicts of interest may not be appropriately
identified or managed. Due to the complexity and opacity of certain
technologies, firms should have robust practices to appropriately
oversee and understand their use and take steps to identify and
appropriately address any associated conflicts of interest. For
example, without appropriate personnel, a firm may not have the ability
to modify the software or may lack the expertise to understand,
monitor, or appropriately update code, limiting the firm's ability to
identify and appropriately address associated conflicts of interest.
Furthermore, if the firm does not understand how the technology
operates--including whether it takes into consideration the firm's
interest and how it can influence investor conduct--the firm may not
fully understand whether, how, or the extent to which it is placing the
firm's interests ahead of investors' interests. As a result of the
complexity and opacity of PDA-like technologies, a firm needs different
and specific practices to evaluate its use of the technology and
recognize the risk of conflicts presented by that use compared to other
practices. Without appropriate oversight and understanding of the
conflicts of interest that could be amplified when the technology is
incorporated into investor-facing interactions, such as design
elements, features, or communications that nudge or prompt certain or
more immediate action by an investor, investor harm can result.
5. Request for Information and Comment
In August 2021, the Commission issued a request for information and
public comment on the use of DEPs by broker-dealers and investment
advisers, as well as the analytical and technological tools and methods
used in connection with these DEPs.\93\ For purposes of the Request,
the Commission defined DEPs broadly to include behavioral prompts,
differential marketing, game-like features, and other design elements
or features designed to engage retail investors.\94\ The Commission
stated that DEPs may be designed to encourage account opening, account
funding and trading, or may be designed solely to increase investor
engagement with investing apps, as there may be value in the number of
investors interacting with the platform, how often they visit, and how
long they stay.\95\ The Request was issued in part to assist the
Commission and its staff in better understanding the market practices
associated with the use of DEPs by firms, facilitate an assessment of
existing regulations and consideration of whether regulatory action may
be needed to further the Commission's mission in connection with firms'
use of DEPs, as well as to provide a forum for market participants
(including investors), and other interested parties to share their
perspectives on the use of DEPs and the related tools and methods,
including potential benefits that DEPs provide to retail investors, as
well as potential investor protection concerns.\96\
---------------------------------------------------------------------------

\93\ See Request, supra note 90.
\94\ See id. at 49067.
\95\ See id. at 49069.
\96\ As noted in the Request, the market practices explored
included: (i) the extent to which firms use DEPs; (ii) the types of
DEPs most frequently used; (iii) the tools and methods used to
develop and implement DEPs; and (iv) information pertaining to
retail investor engagement with DEPs, including any data related to
investor demographics, trading behaviors, and investment
performance. See id. at 49068.
---------------------------------------------------------------------------

The Commission received over 2,300 public comments, including
submissions provided through an online ``feedback flyer'' that
accompanied the Request and was provided to better facilitate responses
from retail investors.\97\ Commenters offered a wide range of
perspectives on broker-dealers' and investment advisers' use of DEPs,
addressing their purpose, providing information on how investors
interact with them, and offering broad reflections on potential
regulatory action. Commenters also provided views on benefits and risks
related to firms' use of DEPs, as well as the AI/machine learning and
behavioral psychology that firms use to develop and deploy DEPs.\98\
---------------------------------------------------------------------------

\97\ The ``Feedback Flyer'' was attached as Appendix A to the
Request and asked individual investors to provide their comments
with regard to online trading or investment platforms, such as
websites and mobile applications, to provide the Commission with a
better understanding of retail investors' experiences on these
platforms. The Feedback Flyer provided 11 different question
prompts, with an array of both multiple choice, and free text
response options whereby respondents could submit relevant comments.
Comments received in response to the Request are available at
<a href="https://www.sec.gov/comments/s7-10-21/s71021.htm">https://www.sec.gov/comments/s7-10-21/s71021.htm</a>.
\98\ See, e.g., Comment Letter from American Securities
Association (Sept. 30, 2021); Comment Letter from Securities
Arbitration Clinic and Professor of Clinical Legal Education, St.
John's University School of Law Securities Arbitration Clinic, (Oct.
1, 2021) (``St. John's Letter''); Comment Letter from Morningstar,
Inc. and Morningstar Investment Management, LLC (Oct. 1, 2021)
(``Morningstar Letter''); Comment Letter from James F. Tierney,
Assistant Professor of Law, University of Nebraska College of Law
(Oct. 1, 2021) (``Tierney Letter''); Pace University Letter; Comment
Letter from Law Office of Simon Kogan, (Oct. 17, 2021) (``Kogan
Letter'').
---------------------------------------------------------------------------

A number of commenters also provided detailed feedback regarding
the potential need for additional action to address the issues
presented by DEPs and their underlying technology. For example,
multiple commenters raised concerns over the risks of harm to investors
if the Commission did not act, and requested that the Commission
interpret existing regulations in a way

[[Page 53970]]

that would apply to most DEPs and/or adopt additional regulations to
address those risks.\99\ Many of these commenters suggested a need to
address the standards of conduct applicable to broker-dealers and
investment advisers when interacting with retail investors through
digital platforms.\100\ Some of these commenters noted that Reg BI does
not apply to firms with a self-directed brokerage business model,
including those that use DEPs \101\ and provided additional suggestions
that the Commission could take to address firms' use of DEPs.\102\
Others provided detailed opinions as to the application of an
investment adviser's fiduciary duty to DEPs.\103\ A significant number
of commenters also addressed other laws and regulations and their
sufficiency, or lack thereof, in their application to DEPs, including
discussion addressing (i) antifraud and general standards of conduct;
\104\ (ii) regulation of advertising, marketing, and communications
with the public; \105\ (iii) compliance and supervision obligations;
\106\ (iv) data privacy and cybersecurity concerns; \107\ (v) customer
onboarding obligations; \108\ (vi) Commission Staff's 2017 Robo-Adviser
Guidance; \109\ and (vii) the Advisers Act recordkeeping rule.\110\
---------------------------------------------------------------------------

\99\ See, e.g., Comment Letter from Scopus Financial Group
(Sept. 20, 2021); Comment Letter from Better Markets, Inc. (Oct. 1,
2021) (``Better Markets Letter''); Comment Letter from Public
Investors Advocate Bar Association (Oct. 1, 2021) (``PIABA
Letter''); Comment Letter from University of Miami School of Law
Investor Rights Clinic et al. (Oct 1, 2021) (``University of Miami
Letter''); Comment Letter from Fidelity Investments (Oct. 1, 2021);
St. John's Letter; Morningstar Letter. We also considered views
received from the SEC's Investor Advisory Committee on ethical
guidelines for artificial intelligence and algorithmic models used
by investment advisers. See Investor Advisory Committee,
Establishment of an Ethical Artificial Intelligence Framework for
Investment Advisors (Apr. 6, 2023), <a href="https://www.sec.gov/files/20230406-iac-letter-ethical-ai.pdf">https://www.sec.gov/files/20230406-iac-letter-ethical-ai.pdf</a>.
\100\ See, e.g., Pace University Letter (``We believe that
retail investors, particularly novice investors, believe that they
are receiving advice or recommendations from DEPs. This includes the
top mover list, analyst ratings, push notifications, and other DEPs
that encourage investment activity. Many of our survey participants
stated that they believe that these DEPs influenced their decision-
making. At the same time, DEPs may also influence investor decision-
making without investors being conscious of it.''); Comment Letter
from North American Securities Administrators Association (Oct. 1,
2021) (``NASAA Letter'') (``To assist with compliance and to protect
investors, the Commission should provide further guidance as to when
DEP-based communications constitute recommendations. However, given
the speed of technology, NASAA suggests that guidance should not be
limited to any particular DEP, but rather should be focused on the
effects of technologies on investor behavior generally.''); Comment
Letter from Fiduciary Insights and Practice Growth Partners (Sept.
30, 2021) (``Aikin/Mindicino Letter'') (``[A]s the complexity and
heterogeneity of wants, needs, and capabilities of the clientele
rises, the sophistication and artificial intelligence and machine
learning (AI/ML) of the DEPs must increase dramatically.
Commensurately, the internal oversight and regulatory guardrails to
assure that customer/client best interests are served must also
increase.''); see also Comment Letter from Morgan Stanley Wealth
Management (Oct. 1, 2021) (``Morgan Stanley Letter'') (while noting
existing protections, stating that ``[s]hould the Commission believe
additional guidance is necessary, we suggest the adoption of
principles-based, technology neutral adjustments to the existing
regulatory regime to address the fast evolving technological
landscape''); Better Markets Letter; University of Miami Letter
(``As the SEC continues its review of standards applicable to
financial professional[s], it is critical to enhance investor
protection in the fast-growing and increasingly harmful digital
platform environment.'').
\101\ See, e.g., Robinhood Letter (``The SEC acknowledged the
benefits of a self-directed model such as Robinhood's in adopting
Reg BI, explicitly stating that Reg BI does not apply to this
model.'').
\102\ See, e.g., Pace University Letter (``DEPs and online
platforms have expanded access to the market to new investors, while
at the same time influencing the decision-making of those
investors--particularly novice investors--in ways that are often in
conflict with their bests interest.''); see also Tierney Letter;
Better Markets Letter; SIFMA Letter; Morningstar Letter; Morgan
Stanley Letter; University of Miami Letter (``Due to the influential
nature of DEPs, the SEC should enhance the Regulation Best Interest
disclosure obligation and conflict of interest obligation by
requiring firms to flag investor trades and/or positions where there
is a likelihood that the firm will act in a manner adverse to the
investor's position and to notify investors of these potential
actions.'').
\103\ See, e.g., IAA Letter (``Some advisers also use various
analytical and technological tools to develop and provide investment
advice, including through online platforms or as part of enhancing
their in-person investment advisory services. Investment advisers
may also engage in DEPs to develop and provide investor education
and related tools.''); see also Comment Letter from Envestnet Asset
Management, Inc. (Oct. 1, 2021) (``Envestnet Letter''); Comment
Letter from Julius Leiman-Carbia, Chief Legal Officer, Wealthfront
Corporation (Oct. 8, 2021) (``Wealthfront Letter''); NASAA Letter;
Aikin/Mindicino Letter; Better Markets Letter; SIFMA Letter;
University of Miami Letter; Morgan Stanley Letter.
\104\ See, e.g., Comment Letter from Jennifer Schulp, Director
of Financial Regulation Studies, Center for Monetary and Financial
Alternatives, CATO Institute (Oct. 1, 2021) (``CATO Institute
Letter''); Comment Letter from Brandon Krieg, CEO, Stash Financial,
Inc. and Stash Investments LLC (Oct. 1, 2021) (``Stash Letter'');
Wealthfront Letter; IAA Letter; Robinhood Letter; SIFMA Letter;
Tierney Letter.
\105\ See, e.g., PIABA Letter; CATO Institute Letter; IAA
Letter.
\106\ See, e.g., Comment Letter from James J. Angel, Ph.D., CFP,
CFA, Associate Professor of Finance, McDonough School of Business,
Georgetown University (Sept. 30, 2021); IAA Letter; Stash Letter;
Aikin/Mindicino Letter; PIABA Letter; CATO Institute Letter.
\107\ See, e.g., NASAA Letter; Envestnet Letter; Kogan Letter.
\108\ See, e.g., University of Miami Letter.
\109\ See, e.g., Comment Letter from Penny Lee, CEO, Financial
Technology Association (Oct. 1, 2021); IAA Letter.
\110\ See, e.g., Comment Letter from Pamela Lewis Marlborough,
Managing Director and Associate General Counsel, Teachers Insurance
and Annuity Association of America (Oct. 1, 2021); SIFMA Letter;
University of Miami Letter.
---------------------------------------------------------------------------

C. Overview of the Proposal

In view of Commission staff observations, our experience
administering our existing rules, the discussion in section 1.B. above
on the development of PDA-like technologies in firm investor
interactions and the unique risks they raise regarding conflicts of
interest, and comments received in response to the Request, we are
proposing to update the regulatory framework to help ensure that firms
are appropriately addressing conflicts of interest associated with the
use of PDA-like technologies. Specifically, we propose that firms
should be required to identify and eliminate, or neutralize the effect
of, certain conflicts of interest associated with their use of PDA-like
technologies because the effects of these conflicts of interest are
contrary to the public interest and the protection of investors.\111\
---------------------------------------------------------------------------

\111\ See infra section II.A.2.e.
---------------------------------------------------------------------------

Proposed rules 15l-2 under the Exchange Act (17 CFR 240.15l-2) and
211(h)(2)-4 under the Advisers Act (17 CFR 275.211(h)(2)-4)
(collectively, the ``proposed conflicts rules'') are designed to
address the conflicts of interest associated with firms' use of PDA-
like technology when engaging in certain investor interactions, and the
proposed rules would do so in a way that aligns with (and in some
respects may satisfy) firms' existing regulatory obligations.\112\
Except as specifically noted, the texts of proposed conflicts rule
applicable to brokers and dealers (17 CFR 240.15l-2) and the proposed
conflicts rule applicable to investment advisers (17 CFR 275.211(h)(2)-
4) would be substantially identical.\113\ The proposed conflicts rules
would only apply where the firm uses defined covered technology--more
specifically, an analytical, technological, or computational function,
algorithm, model, correlation matrix, or similar method or process that
optimizes for, predicts, guides, forecasts, or directs investment-
related behaviors or outcomes in an investor interaction.
---------------------------------------------------------------------------

\112\ See id.
\113\ Citations herein to the ``proposed conflicts rules''
reference each of the proposed conflicts rules as they would be
codified in each location. Citations to a particular section of the
CFR reference only the proposed conflicts rule that would apply to
broker-dealers or to investment advisers, as applicable.
---------------------------------------------------------------------------

The proposal is designed to be sufficiently broad and principles-
based to continue to be applicable as technology develops and to
provide firms with flexibility to develop approaches to their use of
technology consistent with their business model, subject to the over-
arching requirement

[[Page 53971]]

that they need to be sufficient to prevent the firm from placing its
interests ahead of investor interests. The proposal is also designed to
be consistent with the Commission's prior actions regarding
technological innovation.\114\ We note that the staff has also provided
their views on the industry's expanding use of technology in the
context of robo-advisers \115\ and shared examination findings and
risks associated with the use of robo-advisory products,\116\ among
other areas.
---------------------------------------------------------------------------

\114\ Historically, the Commission has reviewed the changing
technology landscape, provided guidance, and if necessary amended
its regulatory framework to protect investors while still allowing
firms' use of technology to innovate and benefit investors. See,
e.g., Use of Electronic Media for Delivery Purposes, Release No.
7233 (Oct. 6, 1995) [60 FR 53458 (Oct. 10, 1995] (providing
Commission views with respect to the use of electronic media for
information delivery under the Securities Act of 1933, the
Securities Exchange Act of 1934, and the Investment Company Act of
1940); Use of Electronic Media by Broker-Dealers, Transfer Agents,
and Investment Advisers for Delivery of Information, Exchange Act
Release No. 37182 (May 9, 1996) [61 FR 24644 (May 15, 1996)] (``1996
Release'') (providing Commission views on electronic delivery of
required information by broker-dealers, transfer agents and
investment advisers); and Use of Electronic Media, Exchange Act
Release No. 42728 (Apr. 28, 2000) [65 FR 25843 (May 4, 2000)]
(``2000 Release'') (providing interpretive guidance on the use of
electronic media to deliver documents on matters such as telephonic
and global consent; issuer liability for website content; and legal
principles that should be considered in conducting online
offerings). In addition, the Commission has amended regulations to
accommodate evolving technologies and changes in the way investors
consume information. See, e.g., Tailored Shareholder Reports for
Mutual Funds and Exchange-Traded Funds; Fee Information in
Investment Company Advertisements, Investment Company Act Release
No. 34731 (Oct. 26, 2022) (87 FR 72758 [Nov. 25, 2022]) (requiring
layered disclosure for funds' shareholder reports and graphical
representations of fund holdings); Investment Adviser Marketing,
Investment Advisers Act Release No. 5653 (Dec. 22, 2020) [86 FR
13024 (Mar. 5, 2021)] (adopting ``principles-based provisions
designed to accommodate the continual evolution and interplay of
technology and advice,'' and providing specific guidance regarding,
among others, the use of social media). Further, the Commission has
amended regulations to expand the use of electronic filing options
by investment advisers and institutional investment managers and
updated recordkeeping requirements to make them adaptable to new
technologies in electronic recordkeeping. See, e.g., Electronic
Submission of Applications for Orders under the Advisers Act and the
Investment Company Act, Confidential Treatment Requests for Filings
on Form 13F, and Form ADV-NR; Amendments to Form 13F, Advisers Act
Release No. 6056 (June 23, 2022) [87 FR 38943 (June 30, 2022)]; see
also Electronic Recordkeeping Requirements for Broker-Dealers,
Security-Based Swap Dealers, and Major Security-Based Swap
Participants, Exchange Act Release No. 96034 (Oct. 12, 2022) [87 FR
66412 (Nov. 3, 2022)] (``Electronic Recordkeeping Release'').
\115\ See Robo-Advisers, Division of Investment Management
Guidance Update No. 2017-02 (Feb. 2017) (``2017 IM Guidance''),
<a href="https://www.sec.gov/investment/im-guidance-2017-02.pdf">https://www.sec.gov/investment/im-guidance-2017-02.pdf</a> (addressing
among other things, presentation of disclosures, provision of
suitable advice, and effective compliance programs).
\116\ See Observations from Examinations of Advisers that
Provide Electronic Investment Advice, Division of Examinations Risk
Alert (Nov. 9, 2021) (``2021 Risk Alert''), <a href="https://www.sec.gov/files/exams-eia-risk-alert.pdf">https://www.sec.gov/files/exams-eia-risk-alert.pdf</a> (noting, ``[n]early all of the
examined advisers received a deficiency letter, with observations
most often noted in the areas of: (1) compliance programs, including
policies, procedures, and testing.'').
---------------------------------------------------------------------------

The proposal draws upon our authority under section 211(h) of the
Advisers Act and section 15(l) of the Exchange Act. The Dodd-Frank Wall
Street Reform and Consumer Protection Act of 2010 (``Dodd-Frank Act'')
added section 211(h)(2) to the Advisers Act and section 15(l)(2) to the
Exchange Act, each of which, among other things, authorizes the
Commission to ``promulgate rules prohibiting or restricting certain
sales practices, conflicts of interest, and compensation schemes for
brokers, dealers, and investment advisers that the Commission deems
contrary to the public interest and the protection of investors.''
\117\
---------------------------------------------------------------------------

\117\ See Section 913 of the Dodd-Frank Wall Street Reform and
Consumer Protection Act, Pub. L. 111-203, 124 Stat. 1376 (2010). As
noted in note 8 to subsection (l), another subsection (l) is set out
after the first subsection (k) of the Exchange Act.
---------------------------------------------------------------------------

The proposal is intended to be technology neutral. We are not
seeking to identify which technologies a firm should or should not use.
Rather, the proposal builds off existing legal standards and, as
discussed throughout the release, is designed to address certain risks
to investors associated with firms' use of certain technology in their
interactions with investors, regardless of which such technology is
used.\118\ The proposal also is designed to permit firms the ability to
employ tools that they believe would address these risks that are
specific to the particular technology they use consistent with the
proposal. The Commission has long acted to protect investors from the
harms arising from conflicts of interests and will continually assess
the harms and revise those protections in light of the evolution of
practices, including with regard to firms' use of technologies. As
discussed in further detail below, conflicts associated with the use of
PDA-like technologies should be eliminated or their effects neutralized
to protect investors from conflicts of interest associated with firms'
use of PDA-like technologies that results in investor interactions that
place the interests of the firm and its associated persons ahead of
investors' interests.
---------------------------------------------------------------------------

\118\ Firms' use of PDA-like technology may also be subject to
other potential legal and contractual restrictions on the ability
for advisers and brokers to collect and use customer information.
See, e.g., 17 CFR part 248, subpart A (Regulation S-P), requiring,
among other things, brokers, dealers, investment companies, and
registered investment advisers to adopt written policies and
procedures for administrative, technical, and physical safeguards to
protect customer records and information.
---------------------------------------------------------------------------

In particular, the proposed conflicts rules would generally require
the following:
<bullet> Elimination, or neutralization of effect of, conflicts of
interest. The proposed conflicts rules would require a firm to (i)
evaluate any use or reasonably foreseeable potential use by the firm or
its associated person \119\ of a covered technology in any investor
interaction to identify any conflict of interest associated with that
use or potential use; \120\ (ii) determine whether any such conflict of
interest places or results in placing the firm's or its associated
person's interest ahead of the interest of investors; and (iii)
eliminate, or neutralize the effect of, those conflicts of interest
that place the firm's or its associated person's interest ahead of the
interest of investors.
---------------------------------------------------------------------------

\119\ As used in this release, the term ``associated person''
means, for investment advisers, a natural person who is a ``person
associated with an investment adviser'' as defined in section
202(a)(17) of the Advisers Act and, for broker-dealers, a natural
person who is an ``associated person of a broker or dealer'' as
defined in section 3(a)(18) of the Exchange Act.
\120\ Covered technology, conflict of interest, investor
interaction are each defined terms under the proposed rules. See
proposed rules 211(h)(2)-4(a) and 15l-2(a); see also infra sections
II.A.1 and II.A.2.c.
---------------------------------------------------------------------------

<bullet> Policies and procedures. The proposed conflicts rules
would require a firm that has any investor interaction using covered
technology to adopt, implement, and, in the case of broker-dealers,
maintain, written policies and procedures reasonably designed to
achieve compliance with the proposed conflicts rules, including (i) a
written description of the process for evaluating any use (or
reasonably foreseeable potential use) of a covered technology in any
investor interaction; (ii) a written description of any material
features of any covered technology used in any investor interaction and
of any conflicts of interest associated with that use; (iii) a written
description of the process for determining whether any conflict of
interest identified pursuant to the proposed conflicts rules results in
an investor interaction that places the interest of the firm or person
associated with the firm ahead of the interests of the investor; (iv) a
written description of the process for determining how to eliminate, or
neutralize the effect of, any conflicts of interest determined pursuant
to the proposed conflicts rules to result in an investor interaction
that

[[Page 53972]]

places the interest of the firm or associated person ahead of the
interests of the investor; and (v) a review and written documentation
of that review, no less frequently than annually, of the adequacy of
the policies and procedures established pursuant to the proposed
conflicts rules and the effectiveness of their implementation as well
as a review of the written descriptions established pursuant to the
proposed conflicts rules.
Proposed amendments to applicable recordkeeping rules, rules 17a-3
and 17a-4 under the Exchange Act and rule 204-2 under the Advisers Act,
would require firms to make and keep books and records related to the
requirements of the proposed conflicts rules. These proposed amendments
are designed to help facilitate the Commission's examination and
enforcement capabilities, including assessing compliance with the
requirements of the proposed conflicts rules.
The proposal is designed to prevent firms' conflicts of interest
from harming investors while allowing continued technological
innovation in the industry.

II. Discussion

A. Proposed Conflicts Rules

1. Scope
The proposed conflicts rules would apply only when a firm uses
covered technology in an investor interaction. The proposed definitions
are designed to identify those conflicts of interest that firms must
evaluate to determine whether they result in investor interactions that
place the firm's interest ahead of investors' interest and must
therefore be eliminated or their effect neutralized.\121\ The proposed
conflicts rules would apply to all broker-dealers and to all investment
advisers registered, or required to be registered, with the Commission.
---------------------------------------------------------------------------

\121\ See supra section I.B.4 (describing existing technologies
that may involve conflicts of interest) and infra section II.A.2.c
(discussing the proposed definition of a conflict of interest).
---------------------------------------------------------------------------

a. Covered Technology
The proposed conflicts rules would define covered technology as an
analytical, technological, or computational function, algorithm, model,
correlation matrix, or similar method or process that optimizes for,
predicts, guides, forecasts, or directs investment-related behaviors or
outcomes.\122\ The proposed definition is designed to capture PDA-like
technologies, such as AI, machine learning, or deep learning
algorithms, neural networks, NLP, or large language models (including
generative pre-trained transformers), as well as other technologies
that make use of historical or real-time data, lookup tables, or
correlation matrices among others.
---------------------------------------------------------------------------

\122\ Proposed conflicts rules at (a).
---------------------------------------------------------------------------

The rate at which these technologies evolve has increased in recent
years and may continue to increase.\123\ Accordingly, the proposed
definition of covered technology is also designed to capture the
variety of technologies and methods that firms currently use as well as
those technologies and methods that may develop over time. The proposed
definition would include widely used and bespoke technologies, future
and existing technologies, sophisticated and relatively simple
technologies, and ones that are both developed or maintained at a firm
or licensed from third parties.\124\
---------------------------------------------------------------------------

\123\ See e.g., Deloitte, Artificial intelligence: The next
frontier for investment management firms (Feb. 5, 2019), <a href="https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html">https://www.deloitte.com/global/en/Industries/financial-services/perspectives/ai-next-frontier-in-investment-management.html</a>
(stating, for example, that ``[f]irms have recognized a new
opportunity to gain direct distribution to investors, benefit from
enhanced efficiencies in servicing small accounts, and offer value-
added services for advisors. This has translated into a wave of
investment activity, with asset managers and intermediaries
acquiring or investing in robo-advice technology.'') See also Bob
Veres and Joel Bruckstein, T3/Inside Information Advisor Software
Survey (Mar. 14, 2023), <a href="https://t3technologyhub.com/wp-content/uploads/2023/03/2023-T3-and-Inside-Information-Software-Survey.pdf">https://t3technologyhub.com/wp-content/uploads/2023/03/2023-T3-and-Inside-Information-Software-Survey.pdf</a>.
\124\ The SEC has proposed a new rule under the Advisers Act to
prohibit registered investment advisers from outsourcing certain
services or functions without first meeting minimum requirements.
See Outsourcing by Investment Advisers, Investment Advisers Act
Release No. 6176; File No. S7-25-22 (Oct. 26, 2022) [87 FR 68816
(Nov. 16, 2022)] (``Proposed Outsourcing Rule''). We encourage
commenters to review that proposal to determine whether it might
affect comments on this proposal.
---------------------------------------------------------------------------

The proposed definition, however, would be limited to those
technologies that optimize for, predict, guide, forecast, or direct
investment-related behaviors or outcomes. The use of these terms in the
proposed conflicts rules is designed to capture a broad range of
actions. This could include providing investment advice or
recommendations, but it also encompasses design elements, features, or
communications that nudge, prompt, cue, solicit, or influence
investment-related behaviors or outcomes from investors. Investment-
related behavior or outcomes can manifest themselves in many forms in
addition to buying, selling, and holding securities, such as an
investor making referrals or increasing trading volume and/or
frequency. This broad proposed definition is designed to help ensure
that, as innovation and technology evolve and firms expand their
reliance on technologies to provide services to, and to interact with,
investors, our rules remain effective in protecting investors from the
harmful impacts of conflicts of interest.
The proposed definition would apply to the use of PDA-like
technologies that analyze investors' behaviors (e.g., spending
patterns, browsing history on the firm's website, updates on social
media) to proactively provide curated research reports on particular
investment products, because the use of such technology has been shown
to guide or influence investment-related behaviors or outcomes.
Similarly, using algorithmic-based tools, such as investment analysis
tools, to provide tailored investment recommendations to investors
would fall under the proposed definition of covered technology because
the use of such tools is directly intended to guide investment-related
behavior. As an additional example, a firm's use of a conditional auto-
encoder model to predict stock returns would be a covered
technology.\125\ Similarly, if a firm utilizes a spreadsheet that
implements financial modeling tools or calculations, such as
correlation matrices, algorithms, or other computational functions, to
reflect historical correlations between economic business cycles and
the market returns of certain asset classes in order to optimize asset
allocation recommendations to investors, the model contained in that
spreadsheet would be a covered technology because the use of such
financial modeling tool is directly intended to guide investment-
related behavior. Likewise, covered technology would include a
commercial off-the-shelf NLP technology that a firm may license to
draft or revise advertisements guiding or directing investors or
prospective investors to use its services.
---------------------------------------------------------------------------

\125\ An autoencoder return model is an unsupervised learning
method that attempts to model a full panel of asset returns using
only the returns themselves as inputs. See generally S. Gu, B.
Kelly, and D. Xiu, Autoencoder Asset Pricing Models (Sept. 30,
2019), <a href="https://www.aqr.com/Insights/Research/Working-Paper/Autoencoder-Asset-Pricing-Models">https://www.aqr.com/Insights/Research/Working-Paper/Autoencoder-Asset-Pricing-Models</a>.
---------------------------------------------------------------------------

The proposed definition, however, would not include technologies
that are designed purely to inform investors, such as a website that
describes the investor's current account balance and past performance
but does not, for example, optimize for or predict future results, or
otherwise guide or direct any investment-related action. Similarly, the
proposed definition also would not include a technology that predicts
whether an investor would be approved for a particular credit card
issued by the firm's affiliate based on other

[[Page 53973]]

information the firm knows about the investor because the use of such
technology does not, and is not intended to, affect an investment-
related behavior or outcome. For the same reason, the use of a firm's
chatbot that employs PDA-like technology to assist investors with basic
customer service support (e.g., password resets or disputing fraudulent
account activity) would not qualify as covered technology under the
proposed definition.
We request comment on all aspects of the definition of covered
technology, including the following items:
1. Is the scope of the proposed definition of a covered technology
sufficiently clear? We intend for the proposed definition to cover PDA-
like technologies; are there ways we could revise the proposed
definition in order to better accomplish this? Are there any
technologies covered by the proposed definition that go beyond PDA-like
technologies and should be excluded? For instance, should the proposed
definition distinguish between different categories of machine learning
algorithms, such as deep learning, supervised learning, unsupervised
learning, and reinforcement learning processes? Do one or more of these
categories present more investor protection concerns related to
conflicts of interest relative to other categories? Would firms be able
to identify what would and would not be a covered technology for
purposes of the proposed rules? If not, what additional clarity would
be beneficial? We have described examples of technologies to which the
definition would or would not apply. Should the definition be revised
to include or specifically exclude such examples?
2. Would the definition adequately include the technology used by
firms that would present the conflicts of interest and resulting risks
to investors that these proposed rules are designed to address? If not,
how should this definition be changed to further the objective of the
proposed conflicts rules? Please explain your answer, including the
extent to which these technologies do or do not present conflicts of
interest risks to investors. Alternatively, do the technologies
included in the proposed definition include technology that does not
typically result in risks to investors that these proposed rules are
designed to address?
3. Is the proposed definition of covered technology appropriately
calibrated to allow for future technological developments? What
adjustments, if any, should the Commission make to help ensure that the
definition of covered technology will remain evergreen despite future
technological advancements? Conversely, what adjustments to the
definition of covered technology, if any, are necessary to avoid
covering those future technological advancements that do not possess
characteristics that the proposed rules are intended to address?
4. The proposed definition of covered technology only applies to
technologies that are used to optimize for, predict, guide, forecast,
or direct investment-related behaviors or outcomes. Do the terms
``optimize for,'' ``predict,'' ``guide,'' ``forecast,'' and ``direct''
appropriately scope the definition? Is it clear what these terms are
intended to capture or would further explanation be helpful? Are there
certain technologies that would fit within one or more of those terms
but which should be outside the scope of the proposed definition?
Alternatively, are there certain technologies that would fall outside
those terms but which should be within the scope of the proposed
definition? If so, should we use additional or different words to
clarify the meaning? For instance, should we include the term
``influence'' in the definition? If so, how would ``influence'' differ
from the terms ``guide'' or ``direct'' in the definition? Should we use
``nudge'' or ``prompt'' in the definition? Alternatively, should we
remove any of the terms in the proposed definition? For instance, are
the terms ``guide'' and ``direct'' redundant or do they express
distinct meanings within the context of the definition? Does ``guide''
capture broader activity than ``direct'' and cause the rule to capture
technologies that should not be in scope? Should the definition be
limited to technologies that direct or influence an investor?
5. Should the proposed definition of covered technology apply to
technologies that are used to optimize for, predict, guide, forecast,
or direct investment-related behaviors or outcomes, directly or
indirectly? Are there certain PDA-like technologies that optimize for,
predict, guide, forecast, or direct investment-related behaviors or
outcomes indirectly that should be covered by this definition? If so,
what are they and why? If the definition did include the term
``indirectly,'' would it include technologies that should not be
covered by the proposed conflicts rules?
6. Should the definition of covered technology not include
technology that is solely meant to inform investors, as proposed?
7. Does the term ``covered technology'' adequately reflect the
definition? Should some other defined term be used, such as ``covered
processes'' or ``covered methods''? Are there any other terms that
should be used?
8. Does the phrase ``investment-related behaviors or outcomes''
sufficiently clarify the intended scope of the rule and which
technologies would not be within the definition? Is it clear what the
phrase ``investment-related behaviors or outcomes'' would capture or
would further explanation be helpful? Are there certain behaviors or
outcomes that may not be ``investment related'' but should nonetheless
be covered by the proposed definition? For instance, should PDA-like
technologies used for back office or administrative functions, such as
trade settlement, the routing of customers' orders, accounting, or
document review and processing, be included in the covered technology
definition? Are commenters aware of any PDA-like technology that is
used for back office functions, such as the routing of customer orders,
that is also used to engage or communicate with investors (i.e., that
involve an investor interaction)? Are there certain investment-related
activities that may not be ``behaviors or outcomes'' that should be
covered by the definition? Is either ``behavior'' or ``outcome''
overbroad, capturing activities beyond those intended by the
definition? Should a different term, such as ``investment-related
covered technology'' be used?
9. Are there aspects of this definition that should be broadened,
narrowed, revised, removed, or added? For instance, should the
definition be limited to the use of predictive data analytics and/or
artificial intelligence that optimizes for, predicts, guides,
forecasts, or directs investment-related behaviors or outcomes?
Alternatively, should we limit the scope of the definition to
technologies that are used to provide investment advice or
recommendations? Should we otherwise limit the scope to technologies
that are used directly by investors? Should we expressly exclude
technologies that are not used by investors but instead are used by
individuals who are associated with a firm and use the technologies in
communicating with investors?
b. Investor Interaction
The proposed conflicts rules include definitions for both
``investor'' and ``investor interaction.'' \126\ For brokers or
dealers, the definition of investor would include a natural person, or
the legal representative of such natural person, who seeks to receive
or receives services primarily for personal, family or

[[Page 53974]]

household purposes. The definition is designed to capture both
prospective and current retail investors.\127\ For investment advisers,
the definition of investor would include a client or prospective
client, and any current or prospective investor in a pooled investment
vehicle advised by the investment adviser.\128\ The use of PDA-like
technology by investment advisers of pooled investment vehicles, such
as algorithmically targeted advertisements that are designed to solicit
investors in a pooled investment vehicle or algorithmically designed
investment strategies in pooled investment vehicles, present the same
investor protection concerns as advisers that use the same or similar
technology to target or advise their advisory clients. Accordingly, we
are proposing to define ``investor'' so that the proposed conflicts
rules would broadly apply both to clients that receive investment
advisory services from an investment adviser and to investors in a
pooled investment vehicle advised by the investment adviser.\129\
---------------------------------------------------------------------------

\126\ See proposed conflict rules at (a).
\127\ See supra note 6. Broker-dealers are subject to regulation
under the Exchange Act and SRO rules, including a number of
obligations that attach when a broker-dealer offers services to a
retail customer, including making recommendations, as well as
general and specific requirements aimed at addressing certain
conflicts of interest. The application of these obligations can vary
depending on a broker-dealer's business lines and activities, as
well as the level of customer sophistication. See Regulation Best
Interest, Exchange Act Release No. 83062 (May 9, 2018) [83 FR 21574
(May 9, 2018)], at 21575 (``Reg BI Proposing Release''); see, e.g.,
FINRA Rule 2210 (applying broker-dealer obligations related to
communications with the public differently to communications
directed to retail versus institutional investors). Here, the focus
of the proposed rules for broker-dealers is on retail investors.
\128\ See proposed rule 211(h)(2)-4(a) (specifying that ``pooled
investment vehicle'' has the same meaning as in 17 CFR 275.206(4)-8,
meaning any investment company as defined in section 3(a) of the
Investment Company Act of 1940 or any company that would be an
investment company under section 3(a) of that Investment Company Act
but for the exclusion provided from that definition by either
section 3(c)(1) or section 3(c)(7) of the Investment Company Act).
\129\ See proposed conflict rules at (a) (defining
``Investor'').
---------------------------------------------------------------------------

The proposed conflicts rules would generally define investor
interaction as engaging or communicating with an investor, including by
exercising discretion with respect to an investor's account; providing
information to an investor; or soliciting an investor.\130\ This
definition would capture a firm's correspondence, dissemination, or
conveyance of information to or solicitation of investors, in any form,
including communications that take place in-person, on websites; via
smartphones, computer applications, chatbots, email messages, and text
messages; and other online or digital tools or platforms. This
definition would include engagement between a firm and an investor's
account, on a discretionary or non-discretionary basis. This definition
would also capture any advertisements, disseminated by or on behalf of
a firm, that offer or promote services or that seek to obtain or retain
one or more investors. The proposed definition is intended to be
sufficiently broad to encompass the wide variety of methods, using
current and future technologies, that firms could use to interact with
investors.\131\
---------------------------------------------------------------------------

\130\ See proposed conflict rules at (a).
\131\ See generally Investment Adviser Marketing Release, supra
note 19 (a recent Commission rule designed to accommodate the
continual evolution of the use of technology in the investment
adviser industry as it relates to advisers marketing their services
to clients and investors).
---------------------------------------------------------------------------

The proposed definition is generally designed to limit the proposed
conflicts rules' scope to a firm's use of covered technology in
interactions with investors. This aspect of the proposed conflicts
rules recognizes that the conflicts associated with the use of covered
technology in investor interactions present a higher risk of harm to
investors than conflicts associated with technologies that are not used
in such interactions. For instance, a firm could utilize covered
technology to analyze historical data and current market data to
identify trends and make predictions related to the firm's intra-day
liquidity needs, peak liquidity demands, and working capital
requirements. A firm could likewise use covered technology to make
investment decisions about its own assets. Similarly, a firm could
implement covered technology for automation of, for example, ``back
office'' processes like the routing of customers' orders \132\ and
accounting and trade settlement. In each of these examples, the use of
covered technology for these processes does not involve an investor
interaction, and therefore would not be subject to the proposed
conflicts rules.
---------------------------------------------------------------------------

\132\ Although routing of customers' orders is not covered by
this proposal, broker-dealers owe their customers a duty of ``best
execution.'' Best execution requires that a broker-dealer seek to
obtain for its customer orders the most favorable terms reasonably
available in the market under the circumstances. See, e.g., Newton
v. Merrill, Lynch, Pierce, Fenner & Smith, 135 F.3d 266, 270 (3d
Cir. 1998). See also Kurz v. Fidelity Management & Research Co., 556
F.3d 639, 640 (7th Cir. 2009); Geman v. SEC, 334 F.3d 1183, 1186
(10th Cir. 2003); see also FINRA Rule 5310 (Best Execution and
Interpositioning). The Commission recently proposed a rule that, if
adopted, would establish through Commission rule a best execution
standard for broker-dealers. See Regulation Best Execution, Exchange
Act Release No. 96496 (Dec. 14, 2022) [88 FR 5440 (Jan. 27, 2023)].
---------------------------------------------------------------------------

In contrast, when a firm's use or potential use of a covered
technology in any investor interaction could involve a conflict of
interest, a firm would be subject to the framework of the proposed
conflicts rules. The proposed definition of investor interaction does
not make any distinctions based on the manner in which an investor or
the investor's account interacts with the covered technology or on the
manner in which the firm uses the technology in the interaction.
Meaning, ``use'' of covered technology in an investor interaction can
occur directly through the use of a covered technology itself (e.g., a
behavioral feature on an online or digital platform that is meant to
prompt, or has the effect of prompting, investors' investment-related
behaviors) or indirectly by firm personnel using the covered technology
and communicating the resulting information gleaned to an investor
(e.g., an email from a broker recommending an investment product when
the broker used PDA-like technology to generate the
recommendation).\133\
---------------------------------------------------------------------------

\133\ To the extent a broker-dealer uses PDA-like technology to
make a recommendation to a retail customer, the broker-dealer would
also be subject to Reg BI and its attendant obligations, including
the Conflict of Interest Obligation, as to the recommendation.
Similarly, an investment adviser making a recommendation to its
client would also be subject to fiduciary obligations that include a
duty of loyalty under which an adviser must eliminate or make full
and fair disclosure of all conflicts of interest. See Fiduciary
Interpretation, supra note 8.
---------------------------------------------------------------------------

Unlike a purely ministerial or back office function, these examples
involve an investment-related communication with an investor and would
be considered an investor interaction under the proposed definition.
Similarly, a firm may use covered technology to provide individual
brokers or advisers with customized insights into an investor's needs
and interests and the broker or adviser may use this information to
supplement their existing knowledge and expertise when making a
suggestion to the investor during an in-person meeting. Such a scenario
would result in the firm using a covered technology in an investor
interaction under the proposed rules. An investor interaction would
also include firms' use of game-like prompts or marketing that
``nudge'' investors to take particular investment-related actions on
digital platforms. In addition, the investor interaction definition
covers solicitations, for example, a firm utilizing covered technology
that scrapes public data, which the firm in turn uses to solicit
clients through broadcast emails.\134\
---------------------------------------------------------------------------

\134\ See infra section II.A.2.e (acknowledging that although a
firm's use of covered technology to solicit investors to open an
account falls under the definition of an investor interaction, it
may not involve a conflict of interest that would require
elimination or neutralization under the proposed conflicts rules).
On the other hand, a conflict of interest may appear if a firm's
chatbot is programmed to solicit only investors that scraped data
show are heavy gamblers, and thus perceived as being more profitable
to the firm as investors that might invest in risky, high-profit
investments that earn the firm more money relative to other
investments.

---------------------------------------------------------------------------

[[Page 53975]]

The proposed definition of investor interaction would include
interactions that have generally been viewed as outside the scope of
``recommendations'' for broker-dealers.\135\ For example, under the
proposed definition, an investor interaction could include: firms' use
of research pages or ``electronic libraries'' to provide investors with
the ability to obtain or request research reports, news, quotes, and
charts from a firm-created website; or firm's use of technologies to
generate emails to investors as part of a firm-run email communication
subscription that investors can sign up for and customize, and which
alerts investors to items such as news affecting the securities in the
investor's portfolio or on the investor's ``watch list.'' \136\
Accordingly, the proposed definition would capture firm communications
that may not rise to the level of a recommendation, yet are nonetheless
designed to, or have the effect of, guiding or directing investors to
take an investment-related action.
---------------------------------------------------------------------------

\135\ See NASD Notice to Members 01-23 (Apr. 2001) (Online
Suitability--Suitability Rules and Online Communications)
(discussing the types of online communications may constitute
``recommendations'' under the NASD suitability rule); Reg BI
Adopting Release, supra note 8, at section II.B.2 (discussing
factors to consider when determining whether a ``recommendation''
has been made by a broker-dealer).
\136\ See NASD Notice to Members 01-23, id.
---------------------------------------------------------------------------

The proposed definition would exclude from the investor interaction
definition interactions solely for purposes of meeting legal or
regulatory obligations.\137\ These interactions are subject to existing
regulatory oversight and/or do not involve the type of conflicts the
proposed rules seek to address. This exclusion would apply to
interactions with an investor for purposes of obligations under any
statute or regulation under Federal or State law, including rules
promulgated by regulatory agencies. For example, the proposed
definition would exclude interactions with investors solely for anti-
money laundering purposes, such as using PDA-like technologies to
identify and track investor activity for the purposes of flagging
suspected fraudulent transactions and requesting identification and
verification of the transaction from an investor (e.g., sending two-
factor authentication messages).\138\ If a firm, however, includes as
part of such an interaction actions that are not reasonably designed to
satisfy its obligations under applicable law (e.g., circulating a link
to a digital platform that includes features designed to prompt
investors to trade along with the annual delivery of Form ADV), and
such additional actions are otherwise within the definition of an
investor interaction, then such action would be considered an investor
interaction for purposes of the proposed conflicts rules.
---------------------------------------------------------------------------

\137\ See proposed conflicts rules at (a).
\138\ The activities covered under this legal and regulatory
obligation exception would qualify as an investor interaction that
uses covered technology absent this exception. However, as a
practical matter, many of these activities would not involve a
firm's use of covered technology under the proposed definition,
because such activities would not involve an analytical,
technological, or computation function, algorithm, model,
correlation matrix, or similar method or process (e.g., delivery of
Form ADV or summary prospectus pursuant to legal obligations).
---------------------------------------------------------------------------

In addition, the proposed definition would also exclude
interactions solely for purposes of providing clerical, ministerial, or
general administrative support. For example, the proposed definition
would exclude basic chatbots or phone trees that firms use to direct
customers to the appropriate customer service representative. This
aspect of the exclusion is only intended to cover basic or first-level
customer support designed to efficiently answer simple questions like
providing the business hours of a branch office or the balance in the
investor's account, or to guide the investor to a human representative
in the appropriate department of the firm who is trained to address the
investor's question. On the other hand, if a firm sought to employ a
more advanced chatbot designed to answer complex investment-related
questions, such as when or whether to invest in a particular investment
product or security, this would no longer fit within the exclusion for
clerical, ministerial, or general administrative support, and would
constitute an investor interaction under the proposed definition.
In either case, the exclusions would be limited to interactions
that are ``solely for the purpose'' of the relevant category (or
categories) of conduct in order to help ensure that interactions that
serve several purposes, including purposes that are not excluded, will
be within the scope of the definition of investor interaction.\139\ The
``solely for the purpose'' language is designed to help ensure that all
the functions of a dual-use technology like a chatbot would be
considered when evaluating conflicts of interest associated with use of
the chatbot.
---------------------------------------------------------------------------

\139\ Interactions that are for the purpose of both categories
of conduct would also fit within the exclusion. For example, an
algorithm whose purpose was both to comply with legal or regulatory
obligations and to conduct other clerical, ministerial, or general
administrative support functions would fit within the exclusion so
long as the algorithm did not also have a third purpose that was not
excluded from the definition.
---------------------------------------------------------------------------

We request comment on all aspects of the proposed definitions of
investor interaction and investor, including the following items:
10. For broker-dealers, the proposed definition of investor means a
natural person, or the legal representative of such natural person, who
seeks to receive or receives services from the broker-dealer primarily
for personal, family or household purposes. Should we narrow the
definition of investor as applied to broker-dealers to only cover
retail customers, as defined under Reg BI? Should we expand the
definition of investor for brokers or dealers to cover all current and
prospective investors and not just retail investors? We have stated
that investors may not be able to understand the complexities of
covered technologies and any conflicts associated with their use.
Should we expand the definition of investor for broker-dealers to cover
a certain subset of non-retail investors? The proposed definition of
investor for investment advisers is not limited to services ``primarily
for personal, family or household purposes.'' Should we add such
limitation in the investment adviser conflicts rule?
11. Should we narrow the definition of investor for investment
advisers? For example, should we only apply it to retail investors, as
defined in Form CRS? If so, please explain why in comparison to other
rules under the Advisers Act.
12. For investment advisers, the proposed definition of investor
also includes investors or prospective investors in a pooled investment
vehicle that is a client or prospective client of the investment
adviser; should we retain this in the final rules? Are there special
considerations for investors in a pooled investment vehicle that cause
them to need less protection from conflicts of interest associated with
a firm's use of covered technology? If the definition of ``investor''
continues to include investors in pooled investment vehicles, as
proposed, are there certain structures or types of pooled investment
vehicles that should not be included? For example, should investors in
collateralized loan obligation vehicles be excluded? Are there unique
characteristics of such vehicles,

[[Page 53976]]

investors, or investors in other pooled investment vehicles, which make
the additional protections that would be provided by the proposed
conflicts rules unnecessary? The proposed definition of ``investor''
would incorporate the definition of ``pooled investment vehicle'' in
rule 206(4)-8. Should we define the term ``pooled investment vehicle''
(or use another term)? Should we define the term more broadly for
purposes of this rule to include other vehicles to which an investment
adviser may provide investment advice that rely on other exclusions
from the definition of investment company, such as companies primarily
engaged in holding mortgages that are excluded pursuant to section
3(c)(5)(C) of the Investment Company Act, or collective investment
trust funds or separate accounts excluded under section 3(c)(11) of the
Investment Company Act?
13. Will the proposed definition of investors present challenges
for firms that are dually registered as investment advisers and broker
dealers?
14. Should we define ``prospective investor'' in the proposed
rules? If so, how should we define this term and why? For example,
should we define ``prospective investor'' as any person or entity that
engages in some way with a firm's services (e.g., downloads the firm's
mobile app, visits the firm's website, or creates a log-in)? If not,
should we provide guidance regarding how firms can identify prospective
investors?
15. Is the proposed definition of investor interaction sufficiently
clear? Would firms be able to identify what would be an investor
interaction for purposes of the proposed conflicts rules? Are there
activities that are not covered by the proposed definition of investor
interaction that should be? Are there activities that are covered by
the proposed definition that should not be? For instance, should a firm
soliciting prospective investors be included within the definition?
Should the proposed definition be limited to interactions in which
investors directly interact with, or otherwise directly use, covered
technology? Do situations in which investors do not directly interact
with covered technology raise the same concerns of scalability as those
in which investors do interact directly?
16. Do commenters agree that investor interactions, as proposed,
may entail conflicts of interest that are particularly likely to result
in investor harm or to take additional effort to discern? Are there
types of activities we should specifically include or exclude within
the definition?
17. Do commenters agree that the definition of investor interaction
should exclude interactions solely for purposes of meeting legal or
regulatory obligations or providing clerical, ministerial, or general
administrative support? Should we remove any or all aspects of these
exclusions from the definition in the final conflicts rules? In the
case of interactions solely for the purpose of meeting legal or
regulatory obligations, should we broaden or narrow the exclusion? For
example, should we take into account legal or regulatory obligations as
a result of compliance with foreign law, or with policies, rules, or
directives of SROs (including securities exchanges) or other bodies?
Generally, would investor interactions that fall under the proposed
exclusions employ covered technology (e.g., technologies that optimize
for, predict, guide, forecast, or direct investment-related behaviors
or outcomes)? If so, how? If not, is the exception for legal or
regulatory obligations additive? Is the exclusion for providing
clerical, ministerial, or general administrative support sufficiently
clear? For instance, is it clear this phrasing would capture trade
settlement and the routing of customers' orders or would further
explanation be helpful?
18. Do the proposed conflicts rules adequately address how a firm
would treat a single covered technology that features functions that
are both included and excluded from the investor interaction
definition? For instance, a chatbot that is used for both general
customer support help (e.g., password resets) and to provide more
advanced functions, such as guiding an investor as to when and whether
to invest in a particular investment product. Should the proposed
conflicts rules treat these dual-purpose covered technologies
differently than covered technology used solely for purposes of meeting
legal or regulatory obligations or providing clerical, ministerial, or
general administrative support?
19. To the extent we retain or expand the exclusions, are there any
conditions we should add in order for a firm to be able to rely on
particular exclusions? For example, should we require that a firm
create and maintain a written record if it relies on an exclusion? Are
there other activities that should be excluded? For example, should we
provide a more principles-based exclusion for certain activities that
the firm affirmatively identifies in writing as low-risk and that are
already part of existing compliance programs or subject to other laws,
rules, regulations, or policies?
20. As specified in the proposed definition of investor
interaction, the definition would include discretionary management of
accounts where the engagement is with the investor's account, even if
there is no communication or other interaction with investors
themselves at the time of trades in their accounts. Should the
discretionary management of accounts be included within the definition
of investor interaction? Should it be excluded? Do commenters agree
that a firm's discretionary management of accounts using covered
technologies may entail conflicts of interest that are particularly
likely to result in investor harm and are not sufficiently addressed
under the current applicable legal framework? Why or why not?
2. Identification, Determination, and Elimination, or Neutralization of
the Effect of, a Conflict of Interest
The proposed conflicts rules would require a firm to eliminate, or
neutralize the effect of, certain conflicts of interest associated with
the use of a covered technology in investor interactions.\140\ The
proposed conflicts rules would also require firms to take affirmative
steps as a precursor to eliminating or neutralizing the effect of these
conflicts. First, a firm would be required to evaluate any use or
reasonably foreseeable potential use of a covered technology in any
investor interaction to identify whether it involves a conflict of
interest, including through testing the technology. Second, a firm
would be required to determine if any such conflict of interest results
in an investor interaction that places the interest of the firm or an
associated person ahead of investors' interests. Third, the proposed
conflicts rules would require a firm to take a particular action--
elimination or neutralization--to address any conflict of interest the
firm determines in step two results in an investor interaction that
places its or an associated person's interest ahead of investors'
interests.\141\ The proposed conflicts rules thus supplement, rather
than supplant, existing regulatory obligations related to conflicts of
interest, laying out particular steps a firm must take to address
conflicts of interest arising specifically from the use of covered
technologies in investor interactions.\142\

[[Page 53977]]

This is because the nature of these technologies (for example due to
their inherent complexity and ability to rapidly scale transmission of
conflicted actions across a firm's investor base) requires additional
steps to address conflicts associated with their use in investor
interactions, compared to conflicts of interest more generally.
---------------------------------------------------------------------------

\140\ See infra section II.A.2.e.
\141\ On the application to interests of associated persons, see
infra sections II.A.2.c, II.A.2.d, and II.A.2.e, and proposed
conflicts rules at (b)(2) and (3).
\142\ The elimination or neutralization requirement of the
proposed rules applies only to a narrower, defined subset of the
broader universe of conflicts--those conflicts that a firm
determines actually place the interests of the firm or certain
associated persons ahead of the interests of investors. This is in
contrast to, for example, an investment adviser's fiduciary duty,
which encompasses any interest that might incline the adviser,
consciously or subconsciously, to provide advice that is not
disinterested., or similarly in contrast to the broader universe of
conflicts covered by Reg BI. Other conflicts of interest that only
might affect the firm's investor interactions would continue to be
subject to these other obligations, as applicable.
---------------------------------------------------------------------------

a. Evaluation and Identification
The proposed conflicts rules would require a firm to evaluate any
use or reasonably foreseeable potential use by the firm or its
associated persons of a covered technology in any investor interaction
to identify any conflict of interest associated with that use or
potential use.\143\ This requirement of the proposal, in connection
with the requirement to test and periodically retest any covered
technology, is designed to help ensure that a firm has a reasonable
understanding of whether its use or reasonably foreseeable potential
use of the covered technology in investor interactions would be
associated with a conflict of interest.
---------------------------------------------------------------------------

\143\ See proposed conflicts rules at (b)(1).
---------------------------------------------------------------------------

The proposed conflicts rules do not mandate a particular means by
which a firm is required to evaluate its particular use or potential
use of a covered technology or identify a conflict of interest
associated with that use or potential use. Instead, the firm may adopt
an approach that is appropriate for its particular use of covered
technology, provided that its evaluation approach is sufficient for the
firm to identify the conflicts of interest that are associated with how
the technology has operated in the past (for example, based on the
firm's experience in testing or based on research the firm conducts
into other firms' experience deploying the technology) and how it could
operate once deployed by the firm. If a technology could be used in a
variety of different scenarios, the firm should consider those
scenarios in which it intends that the technology be used (and for
which it is conducting the identification and evaluation process). It
should also consider other scenarios that are reasonably foreseeable
unless the firm has taken reasonable steps to prevent use of the
technology in scenarios it has not approved (for example, by limiting
the personnel who are able to access the technology).
A firm could adopt different approaches for different covered
technologies.\144\ Such approaches could vary depending on the nature
of the covered technologies employed by the firm at the time they are
implemented, how the technologies are used, and the firm's plans for
future use of those technologies. For example, a firm that only uses
simpler covered technologies in investor interactions, such as basic
financial models contained in spreadsheets or simple investment
algorithms, could take simpler steps to evaluate the technology and
identify any conflicts of interest, such as requiring a review of the
covered technology to confirm whether it weights outcomes based on
factors that are favorable for the adviser or broker-dealer, such as
the revenue generated by a particular course of action.\145\ Even when
a firm identifies a conflict of interest associated with a simple
covered technology, depending on the facts and circumstances, it may
determine that such conflict of interest does not actually result in
the firm's or an associated person's interests being placed ahead of
those of investors, and that the conflict of interest does not need to
be eliminated or its effects to be neutralized.
---------------------------------------------------------------------------

\144\ Cf. U.S Chamber of Commerce Technology Engagement Center,
Report of the Commission on Artificial Intelligence Competitiveness,
Inclusion, and Innovation (Mar. 9, 2023), at 82 (``Chamber of
Commerce AI Report''), <a href="https://www.uschamber.com/assets/documents/CTEC_AICommission2023_Report_v6.pdf">https://www.uschamber.com/assets/documents/CTEC_AICommission2023_Report_v6.pdf</a> (calling for ``impact
assessments'' to help categorize potentially harmful uses of certain
technologies in a risk-based framework).
\145\ See infra section II.A.2.d, discussing financial models.
---------------------------------------------------------------------------

Firms that use more advanced covered technologies may need to take
additional steps to evaluate technology adequately and identify
associated conflicts adequately.\146\ For example, a firm might
instruct firm personnel with sufficient knowledge of both the
applicable programming language and the firm's regulatory obligations
to review the source code of the technology, review documentation
regarding how the technology works, and review the data considered by
the covered technology (as well as how it is weighted).\147\ A firm
seeking to evaluate an especially complex covered technology and
identify conflicts of interest associated with its use may consider
other methods as well. For example, if a firm is concerned that it may
not be possible to determine the specific data points that a covered
technology relied on when it reached a particular conclusion, and how
it weighted the information, the firm could build ``explainability''
features into the technology in order to give the model the capacity to
explain why it reached a particular outcome, recommendation, or
prediction.\148\ By reviewing the output of the explainability
features, the firm may be able to identify whether use of the covered
technology is associated with a conflict of interest.\149\ Developing
this capability would require an understanding of how the model
operates and the types of data used to train it.
---------------------------------------------------------------------------

\146\ These steps could be included in the policies that the
firm would be required to adopt under the proposed conflicts rules,
and may also be necessary to satisfy the proposed recordkeeping
amendments. See infra section II.A.3 and II.B. A written description
of a covered technology prepared in accordance with policies and
procedures that are reasonably designed to prevent violation by the
firm of the proposed conflicts rules generally should include a
written evaluation of the technology and identify any conflicts of
interest presented by the technology. This would also assist the
firm in preparing records that would comply with the proposed
recordkeeping amendments. See infra section II.B.
\147\ When evaluating the data considered by a covered
technology used by a firm, both the data itself and the weighting of
the data may inform a firm's determination of whether or not any
conflict of interest it identifies and evaluates would result in an
investor interaction that places the interest of the firm ahead of
the interests of investors. See infra section II.A.2.d.
\148\ See supra section I.B.4 (describing complex or opaque
technologies, sometimes referred to as ``black boxes'').
\149\ Testing (such as A/B testing) that is designed to
determine the influence of a particular factor may also be helpful
and is discussed infra. If the output of the explainability features
is not sufficient for the firm to identify whether a conflict of
interest exists at all, the firm may still be able to use the output
to determine that any conflict of interest that may exist still does
not result in its interests being placed ahead of investors'
interests, or alternatively that any conflicts of interest that may
exist have been eliminated or their effect has been neutralized due
to controls the firm placed on its use of the technology. See infra
section II.A.2.d (discussing using explainability features for
determination) and infra section II.A.2.e (discussing using
explainability features for elimination or neutralization).
---------------------------------------------------------------------------

Not all of these steps would be necessary (or possible) in all
circumstances. So long as the firm has taken steps that are sufficient
under the circumstances to evaluate its use or reasonably foreseeable
potential use of the covered technology in investor interactions and
identify any conflicts of interest associated with that use or
potential use, this aspect of the proposed conflicts rules would be
satisfied. To the extent a technology is customizable, we anticipate a
firm will be able to evaluate the technology and identify the conflicts
associated with its use through the choices it makes when customizing
the technology. For

[[Page 53978]]

technologies that are not customizable, we anticipate a firm will be
able to evaluate the technology and identify conflicts via other means.
For example, a firm that licenses a covered technology from a third
party may have no access, or limited access, to the underlying source
code of the technology. In such circumstances, provided that the other
documentation regarding how the technology functions is sufficiently
detailed as to how the technology works, the identification and
evaluation could be satisfied through review of such documentation.
Firms without access to the underlying source code could review, for
example, documentation about how the technology can be tailored to its
investors' requirements (such as how to tailor it to eliminate, or
neutralize the effect of, conflicts of interest). In circumstances
where the firm is relying only on the technology's documentation, its
testing methodology would be of special importance to help the firm
discover whether there is any undocumented functionality that could be
associated with a conflict of interest.
When evaluating a covered technology and identifying conflicts of
interest, a firm should consider the circumstances in which a covered
technology would be deployed in investor interactions. Firms that use a
covered technology in investor interactions that operates autonomously
or with limited involvement by firm personnel should consider
subjecting it to more scrutiny because the firm's personnel may not
immediately notice if the conflicts become apparent once the technology
is deployed, or if its outputs change over time.\150\ On the other
hand, if a covered technology is only used to provide first drafts of
marketing materials, or is only used to provide investment ideas that
will be more fully considered by firm personnel who are trained on the
firm's compliance policies, and the drafts or ideas are subjected to
scrutiny throughout the review process before the output is ultimately
used in an investor interaction, the covered technology generally may
need comparatively less scrutiny.
---------------------------------------------------------------------------

\150\ This tendency would also mean that the technology would
need to be tested on a more frequent basis. See infra section
II.A.2.b (discussing proposed testing requirement as it would apply
to technologies that ``drift'' or that operate autonomously).
---------------------------------------------------------------------------

In certain cases, it may be difficult or impossible to evaluate a
particular covered technology or identify any conflict of interest
associated with its use or potential use within the meaning of the
proposed rules. For example, many large language models may consider
millions of different data points, which could make it difficult for a
firm to determine whether certain of those data points implicate the
firm's interest. In some cases, it may be difficult for the firm to
understand exactly what is in the data set that the model is
considering, for example, if it was trained on a data set from the
entire internet. Likewise, there may be situations where a firm does
not have full visibility into all aspects of how a covered technology
functions, such as if the firm licensed it from a third party.\151\
However, a firm's lack of visibility would not absolve it of the
responsibility to use a covered technology in investor interactions in
compliance with the proposed conflicts rules.
---------------------------------------------------------------------------

\151\ FINRA has stated that outsourcing an activity or function
to a third-party vendor does not relieve broker-dealers of their
supervisory obligations, which must be reasonably designed to
achieve compliance with Federal securities laws and regulations, as
well as FINRA rules. See Vendor Management and Outsourcing, FINRA
Regulatory Notice 21-29 (Aug. 13, 2021), <a href="https://www.finra.org/sites/default/files/2021-08/Regulatory-Notice-21-29.pdf">https://www.finra.org/sites/default/files/2021-08/Regulatory-Notice-21-29.pdf</a>. We also
recently proposed a rule that, if adopted, would govern outsourcing
by investment advisers of certain covered functions, and would in
certain cases require investment advisers to obtain reasonable
assurances that third parties could meet certain standards required
by the Advisers Act and the rules thereunder. See Proposed
Outsourcing Rule, supra note 124.
---------------------------------------------------------------------------

The Commission is aware that some more complex covered technologies
lack explainability as to how the technology functions in practice, and
how it reaches its conclusions (e.g., a ``black box'' algorithm where
it is unclear exactly what inputs the technology is relying on and how
it weights them). The proposed conflicts rules would apply to these
covered technologies, and firms would only be able to continue using
them where all requirements of the proposed conflicts rules are met,
including the requirements of the evaluation, identification, testing,
determination, and elimination or neutralization sections. For example,
as a practical matter, firms that use such covered technologies likely
may not meet the requirements of paragraph (b) of the proposed
conflicts rules where they are unable to identify all conflicts of
interest associated with the use of such covered technology. However,
in such cases, firms may be able to modify these technologies, for
example by embedding explainability features into their models and
adopting back-end controls (such as limiting the personnel who can use
a technology or the use cases in which it could be employed) in a
manner that will enable firms to satisfy these requirements.
We request comment on all aspects of the proposed conflict rules'
identification and evaluation requirement, including the following
items:
21. Do the proposed conflicts rules' identification and evaluation
requirements complement, overlap with, or duplicate the existing
regulatory framework for broker-dealers and investment advisers? If so,
in what ways? Specifically, would firms' compliance with those other
regulatory requirements contribute to compliance with the proposed
conflicts rules, and vice versa?
22. Is the proposed requirement that a firm evaluate any use or
reasonably foreseeable potential use of a covered technology to
identify any conflict of interest associated with that use or potential
use sufficient for a firm to understand how it should comply with the
proposed conflicts rules? Should firms only be required to evaluate a
technology used in investor interactions and identify associated
conflicts of interest if they reasonably believe their use (or
potential use) of the technology could be associated with a conflict of
interest that results in their interest being placed ahead of
investors' interests? Absent the evaluation and identification required
under the proposed rule, how would firms form such a reasonable belief?
Should we use some other standard, such as a good faith, recklessness,
or actual knowledge standard, or some other option? Would such a
standard be sufficient to protect investors from the potential harmful
impact of conflicts of interest? Is the requirement sufficiently
general that it would continue to apply to future technologies with
features we may not currently anticipate? If we were to provide
additional clarity (whether through guidance or by changing the
regulatory text), how should we ensure that the rule's requirement to
identify and evaluate these conflicts is sufficiently general that it
would continue to apply to future technologies with features or
functionality that we may not currently anticipate? Should we define
the terms ``identify'' or ``evaluate'' in the regulatory text and, if
so, how should they be defined? Should we use different terms to
address this concept and, if so, which terms and how should they be
defined?
23. The identification and evaluation requirement would also
require firms to identify and evaluate conflicts of interest associated
with use or potential use of a covered technology by an associated
person; what challenges, if

[[Page 53979]]

any, would firms face due to this aspect of the proposed conflicts
rules? Should we make any changes as a result? For example, should we
limit the scope of the requirement to conflicts of interest of which
the firm is aware or reasonably should be aware or should we limit the
scope to any conflict that is reasonably foreseeable? Instead of or in
addition to covering conflicts of interest associated with firms'
associated persons' use of covered technologies, should we prescribe
any additional requirements, such as additional diligence or policies
and procedures, relating to conflicts of interest associated with
firms' associated persons' use of covered technologies? The proposed
conflicts rules would consider conflicts of associated persons only for
associated persons that are individuals, and not of entities that
control, are controlled by, or are under common control with a firm,
but many of the Commission's enforcement actions relating to
undisclosed conflicts have involved conflicts of firms' affiliated
entities, and not of individuals.\152\ In addition to natural persons,
should we broaden the requirement to cover entities controlling,
controlled by, or under common control with firms?
---------------------------------------------------------------------------

\152\ See, e.g., In re. Charles Schwab & Co, supra note 89.
---------------------------------------------------------------------------

24. Do the proposed conflicts rules provide appropriate clarity
around when a firm uses covered technology in an investor interaction?
For instance, is the guidance included in this release clear that the
proposed conflicts rules would not distinguish between a firm directly
using a covered technology in an investor interaction, such as when an
investor interfaces with the covered technology without an intermediary
of the firm, and when a firm uses covered technology indirectly in an
investor interaction, such as where staff of the firm receives the
output and communicates it to the investor? Do commenters agree with
this scope? Should we instead exclude ``indirect'' use in investor
interactions? Alternatively, should we include indirect uses in
investor interactions but apply the rule differently? If so, what
safeguards, if any, would be necessary or appropriate for indirect uses
in investor interactions? As an example, should the rule make a
distinction between an investor interaction using a covered technology
itself (e.g., a behavioral feature on a digital platform) and an
investor interaction in which the firm uses covered technology
indirectly (e.g., a broker emailing a recommendation that it generated
using AI-tools)? Should we revise the rule text to explicitly include
``indirect'' investor interactions, for example by adding the phrase
``directly or indirectly''? Alternatively, should the rule text include
a definition of ``use'' within the context of a firm's use of a covered
technology in an investor interaction?
25. How can scalability rapidly exacerbate the magnitude and
potential effect of the conflict in a way that could make full and fair
disclosure and informed consent unachievable or more difficult? Does
this depend on who the investors are (e.g., individuals versus
entities)? Is it possible to disclose conflicts that are associated
with the use of certain covered technologies in a manner that would
enable investors to understand and provide consent? What are the
characteristics of such technologies, and how do they differ from PDA-
like technologies? How should the final conflicts rules account for
such technologies? For instance, should certain uses of covered
technologies by firms not be subject to the identification,
determination, and elimination or neutralization requirements in the
proposed conflicts rules? Should we permit firms to provide disclosure
regarding their use of such technologies as an alternative method of
complying with the proposed conflicts rules? If so, should the final
rules contain principles pursuant to which firms would decide whether
and how they are able to disclose the conflicts? Should the Commission
instead adopt disclosure standards or criteria? What would those
disclosure standards or criteria entail? For example, should one such
standard be that the technology is easily understandable to laypersons?
What would constitute ``easily understandable to laypersons''?
Alternatively, should the Commission set out different classes of
conflicts of interest or different classes of covered technologies and
prescribe different ways to address each such conflicts or
technologies?
26. Are there particular methods that firms use to identify and
evaluate conflicts of interest that we should discuss in the proposed
conflicts rules? Should we describe particular methods of
identification and evaluation that would comply with the rules? If we
were to address such methods specifically, how would we ensure that the
rule continues to apply to new technologies and new types of investor
interactions as they develop?
27. How widespread is the use of ``black box''-type models
currently? Under existing law, do firms believe that it is possible to
use black box technologies in compliance with the applicable standard
of conduct and, if so, what steps do they take to comply with the
applicable standard of conduct? How will firms using black box
technologies meet the requirements of the proposed conflicts rules?
Will this require significant changes in firms' practices? What
challenges would firms face when identifying and evaluating conflicts
of interest associated with black box technologies, where the outputs
do not always make clear which inputs were relied on, and how those
inputs were weighted? Are there situations where firms are not able
conclusively to identify and evaluate all potential conflicts of
interest associated with a covered technology, including because it is
a black box? How prevalent are these situations? Will they be able to
identify and evaluate whether a firm interest is being considered, or
to determine whether such interest is being placed ahead of the
interests of investors? Instead of or in addition to the proposed
requirements, should we explicitly require that any technologies used
by firms be explainable? Is our understanding correct that firms could
build ``explainability'' features into the technology in order to give
the model the capacity to explain why it reached a particular outcome,
recommendation, or prediction?
28. How will firms conduct conflict of interest identification and
evaluation using personnel who are well-trained on both the inner
workings of covered technologies used in investor interactions and how
to identify common conflicts of interest under the applicable standard
of conduct? Are there other methods firms may use, such as third-party
consultants and, if so, should we explicitly address these other
methods? For example, should we explicitly permit or require a firm to
rely on an analysis prepared by a third party identifying and
evaluating the conflicts of interest that could be associated with a
particular covered technology? If we were to explicitly address third-
party analyses, are there particular situations we should address? For
example, should we permit firms to rely on analyses by developers of
covered technologies that are licensed to firms? What standards would
be necessary in order for a firm to reasonably rely on a third-party
analysis? For example, should a third-party analyst be required to
demonstrate a particular level of expertise, possess a particular
credential, certification, or license, or be independent from the
developer of the technology or the firm relying on the analysis? How
should firms address situations where the underlying source code is not
available

[[Page 53980]]

or is incomplete, or where it is very complex?
29. When firms license covered technologies used in investor
interactions, is the available documentation sufficient for them to
determine whether such technologies present conflicts of interest? Is
review of such documentation sufficient for a firm to identify and
evaluate conflicts of interest?
b. Testing
As part of the identification and evaluation requirement, the
proposed conflicts rules would include a requirement to test each
covered technology prior to its implementation or material
modification, and periodically thereafter, to determine whether the use
of such covered technology is associated with a conflict of
interest.\153\ This obligation would help ensure that conflicts of
interest that may harm investors are identified in light of how the
covered technology actually operates. For example, such testing may
surface additional information that would not be apparent simply from
reviewing the source code or documentation for the covered technology
or the underlying data it uses. It may also surface pre-existing
business practices of a firm where the firm considers firm-favorable
information in its interactions with investors, and the firm's use of
covered technology that replicates such business practices is
associated with a conflict of interest by causing the technology to
consider such firm-favorable information.
---------------------------------------------------------------------------

\153\ Proposed conflicts rules at (b)(1). Testing would only be
required by the proposed conflicts rules as part of the
identification and evaluation prong of the rules. As a practical
matter, some firms that believe they have eliminated, or neutralized
the effect of, conflicts of interest associated with their use of a
covered technology may wish to confirm this through testing. See
infra section II.A.2.e (describing elimination and neutralization).
---------------------------------------------------------------------------

Although the proposed rules would not specify any particular method
of testing or frequency of retesting that the firm must conduct, there
are two specific times testing is required. A firm would be required to
conduct testing prior to the covered technology being implemented.\154\
A firm also would be required to conduct testing before deploying any
``material modification'' of the technology, such as a modification to
add new functionality like expanding the asset classes covered by the
technology. We would not generally view minor modifications, such as
standard software updates, security or other patches, bug fixes, or
minor performance improvements to be a ``material modification.''
During the time that the material modifications are being tested, a
firm could continue to use an older version of the covered technology
if the firm's use of such previous version of the technology complies
with the proposed conflicts rules.
---------------------------------------------------------------------------

\154\ See infra section II.A.2.e for additional information
regarding drift.
---------------------------------------------------------------------------

The proposed requirement to retest a covered technology
periodically does not specify how often retesting would be required. As
a result, a firm also would need to determine how often, and the manner
in which, to retest covered technologies used in investor
interactions.\155\ As with the proposed identification and evaluation
requirement, a firm's testing methodologies and frequencies may vary
depending on the nature and complexity of the covered technologies it
deploys. Relatively simple or easy-to-understand covered technologies
where the risk of a conflict of interest is low could be subject to
similarly simple testing protocols, and such testing could even take
place concurrently with the firm's efforts to identify and evaluate any
conflicts of interest associated with the covered technology. For
example, firms that use relatively straightforward technology may
determine that it is appropriate to expend the majority of their
testing efforts when technology is first implemented (i.e., first
deployed) or when it is substantially modified, and any periodic
testing may focus only on a sampling of the firm's covered
technologies.
---------------------------------------------------------------------------

\155\ Though the policies and procedures requirement of the
proposed conflicts rules would not explicitly require a firm to
specify how often it would retest its covered technologies, as a
practical matter, many firms may find it easier to comply with the
require

[…truncated; see source link]

View on FederalRegister.gov →Plain-text source

Indexed from Federal Register on August 9, 2023.

This is legal information, not legal advice. Laws vary by jurisdiction and change frequently. Always verify current law with official sources and consult a licensed attorney in your jurisdiction for advice on your specific situation.