What a difference a year makes. 

One year ago, we blogged about the FTC’s settlement with Rytr LLC, an AI writing assistant provider alleged to have contributed to generating fake consumer reviews. As one of the FTC’s inaugural AI enforcement actions against a generative tool, Rytr put businesses on notice: turn a blind eye to AI hallucinations and misuse at your own risk. 

On December 22, the FTC reopened and set aside this settlement.

This reversal marks the beginning of what is likely to be a sustained effort to undo the prior administration’s efforts to implement AI governance safeguards. Though it may mitigate some immediate risk for certain AI tool providers and deployers, it likely fuels longer-term regulatory uncertainty as agency independence further erodes and previously settled enforcement precedents grow less reliable. 

Built on an ambitious theory of liability, Rytr may have been an easier target. Zooming out, however, this FTC has set aside precedent at an unprecedented rate, with more reopenings in 2025 than any year prior (based on this author’s count). Coupling this with its deregulatory posture on AI, little should be considered safe. 

Why the Change of Heart?

As recommended by the White House’s July 2025 AI Action Plan, the FTC is now reviewing all enforcement efforts to modify or set aside those that “advance theories of liability that unduly burden AI innovation.” It does not result from this month’s Executive Order, which supplements these efforts to require the FTC to issue a policy statement on the application of the FTC Act’s prohibition on unfair and deceptive acts or practices, expected in the first quarter of 2026. 

The 2024 settlement was agreed to over the objections of the then-minority Republican commissioners (from both Holyoak and Ferguson). Although each dissenter now occupies a significantly different role – one has departed and the other chairs the agency – it seems that their fingerprints are all over the 2025 order. 

• The FTC now appears to largely reject the idea that a neutral tool provider should be liable for how consumers use that tool. An AI-writing assistant, the reasoning goes, should be treated no different than pencils, paper, printers, or other technological “means and instrumentalities.” As noted by the Director of the FTC’s Consumer Protection Bureau, “[c]ondemning a technology or service simply because it potentially could be used in a problematic manner is inconsistent with the law and ordered liberty.” Under this FTC, liability extends only where a business “knows, or has reason to know” the product will be used for unlawful purposes – a standard not satisfied by Rytr.
• The 2025 order aligns with a more restricted, traditional view of unfairness as a theory of liability under Section 5 of the FTC Act. Less focus on the “speculative harms” decried by then-Commissioner Holyoak, and more on facts that can prove a likelihood of substantial injury. Although the original complaint alleged that Rytr’s service could generate detailed reviews “that would almost certainly be false,” it did not plead that users actually posted any draft reviews, which falls short of a practice “likely to cause substantial injury,” and thus would not meet the “unfairness” standard.
• In what presaged this administration’s approach to AI, Commissioner Ferguson in 2024 warned that regulating it “risks strangling a potentially revolutionary technology in its cradle.” While argument for this reversal can be made in good faith, the administration’s and Ferguson’s perspectives on AI regulation, both herein and elsewhere, suggest that the FTC will aggressively prioritize innovation and regulatory restraint. 

What's Next?

A few additional takeaways:

• We may see more reversals in the coming months. Pending legal challenges, the FTC will likely carefully review every AI-related action by the preceding administration, and potentially others broadly construed as unduly burdening AI and innovation.
• Recognizing that industry relies on enforcement for guidance, Chair Ferguson has continually been wary of unnerving businesses with novel theories of liability. By staying well-within the bounds of traditional enforcement, and relying on Congress to address new harms, it appears that his FTC may not venture outside established precedent in the way its predecessor did (unless political directives require otherwise, as was clear from the FTC’s workshop on gender affirming care for minors).
• But maybe not all enforcement is off the table. Ferguson has supported AI enforcement in the past, including the 2024 DoNotPay settlement, which involved a clearer deception theory based on the business’s unsubstantiated claims about its service’s capabilities. This precedent leaves open the question of whether this FTC will favor cases focused on deception rather than the more malleable theory of unfairness.