WASHINGTON, July 8, The Federal Trade Commission is accepting public comments through July 31 on a proposed policy statement that would treat undisclosed steering of AI outputs as deception under Section 5 of the FTC Act. It’s the agency’s clearest move yet to regulate model behavior as a consumer-protection matter rather than a technical one.

The statement, published July 7 in the Federal Register as FR Doc. 2026-13628, was issued pursuant to Executive Order 14365 and cleared on a 2-0 Commission vote. “The FTC wants to hear from businesses and consumers about their experiences and concerns regarding the subversion of AI systems for ideological ends,” said Chairman Andrew N. Ferguson.

The legal theory, as parsed by Covington’s Inside Privacy analysis, is that vendors make explicit and implicit representations that their systems produce accurate, objective outputs. Configuring a model to pursue undisclosed objectives, whether ideological or in service of state-law compliance, could count as a material deceptive omission. Buchanan Ingersoll & Rooney notes the FTC singled out Colorado’s Artificial Intelligence Act and said following a state statute isn’t a defense to a Section 5 claim. Liability reaches any company that promotes, licenses, or makes accuracy claims about a system. Hallucinations from technical limits are treated separately.

The timing is the story. Intuit’s 2026 AI Impact Report, drawing on more than 34,000 small and midsize business owners and 5.3 million QuickBooks accounts, finds 77 percent of U.S. businesses now use AI regularly, up from 48 percent in July 2024. Marketing, customer service, and bookkeeping lead the use cases. Roughly one in ten firms in each surveyed market pay for dedicated tools, including no-code platforms like Glean, Dust, and LemonLime.

Regulators are meeting the technology at the exact moment it stopped being optional.

Sources