Securities and Exchange Commission Brings First Enforcement Actions Over “AI-Washing”

On March 18, 2024 the US Securities and Exchange Commission (“SEC” or the “Commission”) announced that it had settled charges in separate actions against two investment advisers, Delphia (USA) Inc. (“Delphia”) and Global Predictions, Inc. (“Global Predictions”) for making false or misleading statements regarding their use of artificial intelligence (“AI”). These enforcement actions are the SEC’s first explicit AI-related actions against investment advisers.

The SEC charged both advisers with violations of Section 206(2) and Section 206(4) of the Investment Advisers Act of 1940 (the “Advisers Act”). In addition, the Commission charged both advisers with violations of Investment Advisers Act Rule 206(4)-1 (the “Marketing Rule”), for disseminating advertisements that included untrue statements of material fact or otherwise omitted material facts necessary to make the advertisements not misleading under the circumstances.¹ The Commission also charged the advisers with violations of Advisers Act Rule 206(4)-7 (the “Compliance Rule”), for failure to implement written policies and procedures reasonably designed to prevent the Advisers Act and Marketing Rule violations. The SEC ordered Delphia and Global Predictions to pay civil penalties of $225,000 and $175,000, respectively.

The interest of the SEC and its staff in AI, or “predictive data analytics,” has been growing for some time. In late 2023, the Commission began an AI sweep, requesting information from various investment advisers on AI-related topics. These inquiries correlated with the Commission’s 2024 Examination Priorities report, which states that a main focus will be on emerging financial technology, including AI. In addition, on July 26, 2023, the SEC proposed an Advisers Act rule that would specifically require SEC-registered investment advisers to eliminate or neutralize the effect of conflicts of interest associated with their use of covered predictive data analytics in providing advisory services to clients, and adopt written compliance policies and procedures regarding the same. To date, that rule has not been adopted. The enforcement actions against Delphia and Global Predictions, however, show that the Commission is ready, willing and able to use current federal securities laws to address at least certain of the regulator’s concerns regarding the use of AI by investment advisers.

Ultimately, the SEC’s concerns are nothing new, although the technology is. In addition to consistent regulatory interest in conflicts of interest, the SEC and its staff historically have scrutinized whether investment advisers are managing client assets in a manner consistent with their advertisements and disclosures, particularly where the claimed strategies are new and otherwise trendy. Notable examples in recent years include heightened regulatory interest—including enforcement actions—in investment advisers’ claims regarding ESG (including so-called “green-washing”), “robo,” quantitative and similar algorithmic-based strategies and services.² In each case, the bottom line for the Commission and its staff was that the advisers’ statements to investors and clients about their strategies were not consistent with the manner in which the advisers were actually managing their assets. Whether it is AI-washing, green-washing or quant-washing, the regulatory message is the same—say what you do and do what you say, as further demonstrated by these two recent enforcement actions, which are summarized below.

Delphia

Delphia is a Toronto-based investment adviser that offers robo-adviser portfolio management services to retail clients. In August 2019, Delphia began claiming in its Form ADV, Part 2A brochure, on its website and in a press release that it used AI (including machine learning) in its advisory algorithms and that it also collected client data and incorporated it into these algorithms. For example, the press release stated, among other things, that Delphia used machine learning to “analyze the collective data shared by its members to make intelligent investment decisions.” In another example, from November 2020 to August 2023, Delphia’s website stated that the adviser puts client data to work to make its AI smarter, “so it can predict which companies and trends are about to make it big and invest in them before anyone else.”

During an examination, the SEC’s Division of Examinations “discovered” the various statements described, and in response to SEC staff questions, Delphia admitted to the staff in July 2021 that it had not used any client data, and that it did not have an algorithm to use client data. Shortly thereafter, in August, Delphia updated its brochure to reflect that it did not in fact use client data to make investment decisions.³ Delphia also informed the staff in October 2021 that it would review all current marketing and regulatory disclosure documents, and take action to correct any false and misleading statements regarding the use of client data. Delphia also created the role of Compliance Manager for its compliance team and retained two outside compliance consulting firms. Additionally, in communications with certain investors, Delphia noted that client data was not being used as a data source for its algorithms because it had not yet collected enough client data to provide meaningful insights.

Nonetheless, through August 2023, Delphia continued to make statements to investors in emails and press releases claiming to incorporate client data into its algorithms to make investment decisions. For example, a November 2022 press release stated that “Delphia’s proprietary algorithms combine the data invested by its members with commercially available data, to make predictions across thousands of publicly traded companies up to two years into the future.” In emails to new clients during 2021 and 2022, Delphia stated that the clients’ data was “helping [Delphia] train [its] algorithm for pursuing ever better returns” and that Delphia “will pool your data with everyone else’s to power our algorithm.” In addition, a 2022 social media post (which remained until 2023) said that Delphia’s “proprietary algorithm uses the data being invested by our members, so we can make stock selections across thousands of publicly traded companies up to seven financial quarters in the future.” While Delphia did collect some client data intermittently between 2019 and 2023, it never incorporated it into any AI or machine learning technology or otherwise used that data in any way as inputs into its investing algorithms. The SEC found these additional statements, like the ones noted in the 2021 examination, to be false and misleading because Delphia simply did not possess the capabilities it claimed to have.

In terms of compliance program failures, the SEC said that while Delphia did have “some” advertising policies and procedures, it lacked policies and procedures relating to social media. Further, Delphia had a number of employees and consultants involved in their advertising review-and-approval process. Delphia failed to lay out a clear advertising review and approval process, either in its policies and procedures or otherwise, that would enable its personnel and consultants to understand their respective roles and responsibilities in that process. The SEC thought that this was the reason that, even though Delphia corrected certain false and misleading statements regarding the use of client data after the SEC examination, Delphia continued to misrepresent its use of client data in other communications.

The SEC imposed a civil penalty of $225,000, noting Delphia’s cooperation in this matter. In addition to Delphia’s AI and data usage claims, the regulatory interest of the SEC and its staff in Delphia was likely also heightened because, among other things, Delphia’s clients were retail clients, the advisory service was a “robo” service,⁴ and Delphia made promises to the examination staff that were not fulfilled.

Global Predictions⁵

Similar to Delphia, Global Predictions, a San Francisco-based internet investment adviser, offered advisory services to retail clients through an interactive online platform, which makes investment allocation recommendations to clients. The adviser claimed that the platform made non-discretionary recommendations utilizing algorithms. The platform allows clients to interface with a chatbot that communicates investment recommendations but the chatbot does not itself generate the recommendations.

From its date of registration on August 14, 2023, Global Predictions disseminated advertisements claiming use of AI technologies. For example, the adviser stated on its website that its technology “incorporated [e]xpert AI driven forecasts,” even though it did not. In client emails and on social media posts and websites, Global Predictions claimed to be the “first regulated AI financial advisor,” but could not produce documents to substantiate this claim.

The SEC imposed a $175,000 civil penalty, and recognized the adviser’s cooperation as well as the remedial steps it had taken, including, in relevant part, its removal of violative advertisements, retention of a compliance consultant to review its marketing materials, and compliance training undertaken by its employees. Similar to Delphia, in addition to Global Predictions’ AI claims, the regulatory interest of the SEC and its staff in Global Predictions was likely also heightened because, among other things, its clients were retail clients and the advisory service at issue was a “robo” advisory service.

Conclusion

While the Delphia and Global Predictions settlements are certainly attention-grabbing as the SEC’s first cases on “AI-washing,” investment advisers may be somewhat comforted in the understanding that—at least in these instances—there is no new “rulemaking by enforcement” at work here, but more simply the tried-and-true axiom of “say what you do, and do what you say” applied in a new context.

That said, investment advisers can and should use this as a reminder—and an opportunity—to review and hone statements regarding an adviser’s use of technology in the provision of its advisory services. Advisers are certainly not immune to broader commercial and economic trends, and while the AI boom may have more staying power than the blockchain craze from just a few years ago (or the dot-com boom from longer still), advisers are subject to stricter “truth in advertising” obligations than many other market participants. Advisers should scrutinize marketing materials and other communications that contain terminology relating to AI (e.g., “machine learning,” “deep learning,” “neural networks,” “large language models,” “natural language processing,” or “generative pre-trained transformers” (“GPT”)) or the use of quantitative or algorithmic investment strategies. Further, investment advisers should take steps to ensure that these communications accurately describe the adviser’s actual investment strategy and related process, and appropriately disclose conflicts of interest and material risks associated with the adviser’s use of that technology (particularly where third-party vendors/licensors are involved).

As with almost anything related to marketing (or anything else an investment adviser does in a fiduciary capacity), in order for compliance and legal personnel to implement reasonably designed policies, procedures and internal controls—and to effectively and substantively review the material and identify and address risks and conflicts—they must truly understand both the technology and the manner in which the adviser uses it. Thus, it is crucial that compliance and legal personnel become educated on both of these topics, and continue that education as technology, and the adviser’s use of it, changes over time.⁶

¹ See also SEC Sweep into Marketing Rule Violations Results in Charges Against Nine Investment Advisers; SEC Charges Five Investment Advisers for Marketing Rule Violations; and US SEC Brings First "Marketing Rule" Action: A Return To Rulemaking By Enforcement?.

² See, e.g., 2011 Quant Adviser Enforcement Action, 2018 Quant Related Enforcement Action, 2023 ESG Enforcement Action, 2022 Robo-Adviser Enforcement Action, SEC Charges Robo-Adviser with Misleading Clients, February 2017 Guidance Update, Investor Bulletin: Robo-Advisers, and Observations from Examinations of Advisers that Provide Electronic Investment Advice.

³ However, as the SEC noted, Delphia did not identify the new disclosures in its summary of material changes section of the brochure even though the change related to a core part of its retail advisory program.

⁴ See also footnote 2 above.

⁵ Although this discussion focuses on AI matters, the SEC’s statements in this enforcement action regarding the Marketing Rule, as well as regarding certain terms and conditions that the adviser had included in its investment advisory contracts with retail investors, are very instructive. Regarding marketing for example, the adviser represented on its public website that it offered tax-loss harvesting services that could save users “thousands of dollars,” when it did not in fact offer any tax-loss harvesting services; on its public website and in a press release, the adviser claimed that it had more than $6 billion of assets on its platform, although it does not have or report any regulatory assets under management on its Form ADV; included hypothetical performance on its public website (rather than to a particular intended audience) without providing any disclosures along those lines (see also footnote 4 above); and included testimonials on its public website without disclosing material conflicts of interest (e.g., two persons giving testimonials had outside business relationships with the adviser’s Chief Executive Officer, one of which had previously been retained by the adviser as an independent contractor, and a third person giving a testimonial was a close family member of the Chief Executive Officer). The retail client advisory contracts permitted the adviser to change the terms of the contract unilaterally without advance notice to clients, and required clients to periodically visit the adviser’s website and review the contract themselves for any changes (in fact, after the adviser changed its contract, it sent a form email to clients informing them that the contract had changed, with a link to the updated terms of service, but without identifying what changes had been made or providing a mechanism for clients to provide or withhold informed consent to the change prior to it becoming effective); included statements that were inconsistent with other representations (e.g., the contract inaccurately stated that the adviser “do[es] not give financial or investment advice or advocate the purchase or sale of any security or investment.”); and included a “hedge clause” that purported to relieve the adviser from liability for “any claim or demand”(regardless of the theory of liability) and cause the client to broadly indemnify and hold the adviser harmless from any third-party claim or demand arising out of the client’s use of the adviser’s services (see also Commission Interpretation Regarding Standard of Conduct for Investment Advisers).

⁶See also a recent speech, “Remarks at Program on Corporate Compliance and Enforcement Spring Conference 2024,” from the SEC’s Director of the Division of Enforcement, echoing many of the above sentiments.