�鶹���

Litigation

Advisers settle allegations of falsely representing AI-driven investment advice

On March 18, 2024, the SEC issued two orders instituting and settling administrative and cease-and-desist proceedings against two investment advisers, (Delphia) and (Global Predictions).  Each adviser told potential clients that it would make investment recommendations using artificial intelligence when, in fact, neither allegedly developed the promised artificial intelligence algorithm nor used such an algorithm to provide investment advice.

Global Predictions is a San Francisco-based registered internet investment adviser, and Delphia is a Toronto-based corporation that had been registered as an investment adviser from September 2019 through January 2024.

Starting in 2019, Delphia allegedly sought to use artificial intelligence and machine learning to collect data from its clients (including social media, banking, credit card, and online purchase data) to make investment recommendations, but was not ultimately successful in using that data in an artificial intelligence or machine learning process to make investment recommendations.  Nonetheless, according to the SEC, from 2019 through 2023 Delphia repeatedly told investors that its advice was generated by using personal information in a “predictive algorithmic model” and that it was the “first investment adviser to convert personal data into a renewable source of investable capital” using machine learning.  It also represented that Delphia’s technology “turns your data into an unfair investing advantage.”  Each of these statements was allegedly false because Delphia did not use client data to make investment recommendations.

In a 2021 exam, Delphia admitted to SEC staff that it had not used client data to make recommendations, and had not developed an algorithm to use client data to make investment recommendations.  Delphia took some corrective measures, but continued to represent to investors that it was using client data to “power our algorithm” and that it used “proprietary algorithms” to “combine the data invested by its members … to make predictions” despite not having developed such algorithms.

Similarly, Global Predictions represented to potential clients that it used “expert AI-driven forecasts” to generate investment advice, and that it was the “first regulated AI financial advisor,” but allegedly did not use AI to make investment recommendations.  Global Predictions also allegedly made a number of false or misleading statements about its advice, such as that it had more than $6 billion in assets on its platform despite reporting no assets under management in its Form ADV, that its “models are outperforming IMF forecasts by 34%,” without identifying what the 34% figure referred to and without maintaining back-up documentation to support this claim, and that Global Predictions offered tax-loss harvesting services that could save users “thousands of dollars” when it did not in fact offer any tax-loss harvesting services.   According to the SEC, Global Predictions also presented hypothetical performance in an advertisement to a mass audience rather than presenting hypothetical performance relevant to the likely financial situation and investment objectives of an intended audience, among other violations of the marketing rule under the Advisers Act.

In addition, Global Predictions’ advisory contract with retail clients also included broad liability disclaimer language which the SEC alleges created the misleading impression that clients had waived non-waivable causes of action against Global Predictions provided by state or federal law.

Both Global Predictions and Delphia are also alleged to have failed to implement policies and procedures adequate to prevent the alleged misconduct.

On account of these allegations, Global Predictions and Delphia both agreed to be censured and to cease and desist from further violations.  Global Predictions agreed to pay a civil money penalty of $175,000; Delphia agreed to pay a civil money penalty of $225,000.

While the alleged use of AI-driven investment advice makes these two enforcement actions novel, the core issue could not be simpler:  an adviser may not tell clients that it uses an AI model to make investment recommendations when it does not use an AI model to make investment recommendations.