SEC intensifies scrutiny on ‘AI washing’

The US securities regulator has warned firms not to falsely advertise the use of artificial intelligence in their investment products, adding those that engage in ‘AI washing’ may face enforcement action.

“It is a nice thing to advertise to show investors that ‘hey, we are cutting-edge’, but if the firm is not actually using it, it can attract the attention of enforcement,” said Maurya Keating, associate regional director for the investment adviser and investment company examination programme at the Securities and Exchange Commission.

Keating said her team has observed increased advertising claims relating to the technology as AI becomes a powerful buzzword across the industry, and sales teams seek to capitalise on investor trends. She adds the regulator has engaged in various examinations of companies’ marketing activities to ensure all relevant information is properly disclosed.

Thomas Smith, associate regional director at the SEC Enforcement Division, said firms that refer to AI in their marketing should be ready to show regulators exactly how they are using these tools, in the same way they must be able to show how they are selling the products to investors and customers.

Keating and Smith were speaking at a New York conference hosted by Compliance Week on June 11.

The warnings come as banks step up their use of AI and machine learning tools for building investment products, including in their quantitative investment strategy divisions.

Heightened scrutiny

SEC chair Gary Gensler first signalled the agency’s concerns about AI washing last December, when he warned against the practice: “Don’t do it . . . I don't know how else to say it,” he said.

The following month, the SEC issued an alert urging investors to be aware of increasing investment frauds involving the purported use of such technology.

On March 18, it stepped up the battle when it fined two investment advisers – Delphia and Global Predictions – in the first such actions.

The SEC said Delphia falsely claimed it had been employing AI and machine learning tools to analyse client data and guide investment decisions. In fact, neither the data input nor the AI model usage ever took place. The company made these misleading statements across regulatory filings, press releases and on its website.

If [a] firm is not actually using AI, it can attract the attention of enforcement

Maurya Keating, SEC

Likewise, Global Predictions falsely claimed itself to be the “first regulated AI financial advisor” and misrepresented that its platform provided “expert AI-driven forecasts” when that was not the case.

Smith stated this regulatory focus is not new, and likened it to greenwashing, where companies have been fined for making false claims about their environmental impact. He noted the current scrutiny of AI use aligns with the SEC’s long-standing practice of ensuring companies are honest about their products and services to avoid misleading customers or harming investors.

“Whether [companies] use AI for a core product or a minor piece of the business that only a few people use, as long as it's properly disclosed, and the scope is not overblown or exaggerated, then it probably will draw less questions [from regulators],” says Smith.

Delphia and Global Predictions were fined $225,000 and $175,000, respectively.

The first AI washing enforcements imposed relatively small fines compared with some penalties it has issued for greenwashing. In 2023, the regulator ordered Deutsche Bank subsidiary DWS to pay $19 million relating to mis-statements in the asset management unit’s environmental, social and governance processes.

Even so, the enforcement represents one of many ways in which US regulators have intensified their AI oversight following last October’s presidential directive, which urged financial authorities to manage the risks that come with the new technologies.

Know your products

For companies that have marketed or want to market their use of AI, their compliance teams must fully understand the real capabilities of the tools deployed, and review advertising materials to ensure accuracy.

Speaking at the same event, Matthew Siano, former managing director and general counsel at Two Sigma Investments said this is “not necessarily a skill set that many typical standard legal compliance folks have”.

Suzan Rose, senior adviser on government and regulatory affairs at the Alternative Investment Management Association, raised concerns for smaller and less tech-savvy firms, which directly purchase AI products from third-party vendors and may lack the expertise and resources to fully understand how the AI tools function.

Speaking at the same event, Rose warned these firms could be misled by AI vendors and unintentionally publish misleading marketing information to their investors and customers.

Siano echoed that many firms are facing this challenge. “If you are going to spend your money, your client's money, you better understand that,” he said.

If firms are uncertain about some capabilities of their AI tools, they should not advertise them on websites or social media, he added. “Otherwise, from the investment management side, I think this gives the industry a bad name.”

Some products incorporating AI references in their titles already play down the use of these techniques in the small print.

For example, Credit Suisse’s RavenPack Artificial Intelligence index, which uses natural language processing as a sentiment indicator, notes the reference to AI “does not imply any form of actual intelligence, learning or comprehension” on the front page of its website. The index is sold in fixed indexed annuities, which are offered in insurance wrappers, so sit outside of the SEC’s gaze.