On July 26, 2023, the Securities and Exchange Commission (“SEC”) proposed new rules targeting the use of predictive data analytics and artificial intelligence (“AI”) by registered investment advisers (“RIAs”) and broker-dealers. The new proposed rules focus on the potential for conflicts of interest and the possibility that newer, more complex analytics models (including those using AI) might optimize decision making for RIAs and broker-dealers by placing those firms’ interests above the interests of their clients. The proposed rules would require RIAs and broker-dealers to: (i) evaluate whether their use of technologies “that optimize for, predict, forecast or direct investment-related behaviors or outcomes” create such a conflict of interest, and (ii) either stop using or address the effects of tools that place a firm’s interests before the interests of clients. RIAs and broker-dealers will also will be required to adopt policies to ensure compliance with the new proposed rules.
The SEC identifies several example conflicts of interests in the proposal, including:
- predictive models encouraging clients to keep assets in an advisory account instead of other accounts like a 401(k) or retirement account;
- transaction-based incentives that could encourage predictive models to maximize the frequency of transactions;
- revenue sharing for investment products creating incentives for advisers to favor those investments; and
- AI software promoting a firm’s proprietary products over other products (regardless of the best interests of the customer).
The underlying potential conflicts have long existed, but the SEC appears concerned that the scalability and data-processing capabilities of new technologies could accelerate conflicted transactions in the absence of oversight and regulation. And while the SEC acknowledged that when these technologies are optimized for investor interests real benefits can occur in market access, efficiency and returns, they also noted that investor harm can occur if this technology is optimized to place RIA or broker-dealer interests ahead of investors.
Some more advanced predictive tools are considered “black boxes” because even their creators, let alone users, find it difficult to fully explain how these tools reach their conclusions. The proposed rules cover such “black box” algorithms and the SEC suggested that as a practical matter, these algorithms likely would not satisfy the new rules unless users are able to identify all potential conflicts of interest.
For the more complex tools with millions of potential outputs, it may prove difficult or impossible for regulated firms to precisely identify and evaluate every theoretical conflict of interest. In his dissent from the rule proposal, Commissioner Uyeda expressed concern that the “regulatory vagueness and considerable compliance challenges may cause [RIAs and broker-dealers] to avoid innovation or efficiencies through automation.” And indeed, the SEC acknowledged that one consequence of the proposal could be that “a firm might opt not to use an automated investment advice technology because of the costs associated with complying with the proposed rules.”
- Further regulations are on the way. The government, and particularly the SEC, continues to signal that further regulation related to the use of artificial intelligence is on the way. The proposed rules come just one week after Chair Gensler warned reporters about the risks posed by artificial intelligence through conflicts of interest, bias promotion, financial consolidation and fraud.
- RIAs and Broker-Dealers Who Employ Algorithmic Trading Technology Should Evaluate Their Risk. RIAs and broker-dealers would be wise to take stock of their current or planned AI tools and potentially model whether they could run afoul of the conflict examples identified in the proposing release.
- As Chair Gensler said last week, “fraud is [still] fraud.” As the use of AI continues to proliferate, regulators will find themselves combatting the classic concerns of securities regulation reinvented by AI products.
- The more complex the tool, the more difficult it will be to prove the absence of conflicts of interest. The SEC will need to strike a careful balance between regulating risks and stifling innovation by discouraging investment firms from using AI tools that might lead to unpredictable results before they are perfected.
 SEC Proposes New Requirements to Address Risks to Investors From Conflicts of Interest Associated With the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers, Securities and Exchange Commission (July 26, 2023) https://www.sec.gov/news/press-release/2023-140.
 Conflicts of Interest Associated with the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisors, Securities and Exchange Commission (July 26, 2023) at *8. https://www.sec.gov/files/rules/proposed/2023/34-97990.pdf.
 Id. at *67-68. (“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.”)
 Statement on the Proposals re: Conflicts of Interest Associated with the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisers, Mark Uyeda, Commissioner, Securities and Exchange Commission (July 26, 2023), https://www.sec.gov/news/statement/uyeda-statement-predictive-data-analytics-072623.
 Conflicts of Interest Associated with the Use of Predictive Data Analytics by Broker-Dealers and Investment Advisors, Securities and Exchange Commission (July 26, 2023) at *192.
 “Isaac Newton to AI” Remarks before the National Press Club, Securities and Exchange Commission (July 17, 2023) https://www.sec.gov/news/speech/gensler-isaac-newton-ai-remarks-07-17-2023.