• by Wendy Davis  @wendyndavis, January 10, 2024

The Federal Trade Commission this week warned artificial intelligence companies to honor their privacy promises, including statements made in promotional materials and terms of service.

In a Tuesday blog post, the agency said “model-as-a-service” companies -- meaning artificial intelligence companies that train large language models, then sell access to stores, banks and other businesses that deploy chatbots -- may face enforcement actions if they renege on privacy promises.

“There is no AI exemption from the laws on the books,” the FTC wrote. “Like all firms, model-as-a-service companies that deceive customers or users about how their data is collected -- whether explicitly or implicitly, by inclusion or by omission -- may be violating the law.”

The agency elaborated that it could bring enforcement actions against companies that violate promises to refrain from using customer data to train or update their models. The FTC noted it recently brought cases against GoodRx and BetterHelp for allegedly disclosing people's data for ad targeting after promising to keep the information confidential.

“What a company fails to disclose to customers may be just as significant as what it promises,” the agency wrote, adding companies that “omit material facts” -- including how they collect and use customer data -- could face enforcement actions.

“Model-as-a-service companies must also abide by their commitments to customers regardless of how or where the commitment was made,” the FTC added. “This includes, for instance, commitments made through promotional materials, terms of service on the company’s website, or online marketplaces.”

The agency additionally warned artificial intelligence companies against making key disclosures in “behind hyperlinks, in legalese, or in fine print.”