On December 22, 2025, the FTC reopened and set aside its consent order against Rytr, the AI-writing startup, citing America’s AI Action Plan. The commission said continued enforcement of the Rytr order would “chill legitimate AI innovation.” Translation: the Trump FTC thinks some of the previous administration’s AI enforcement crossed from protection into discouragement.
This looks like retreat. It’s selective retreat. Five Operation AI Comply orders from September 2024 — DoNotPay, Ascend Ecom, Ecommerce Empire Builders, FBA Machine, IntelliVision — remain in force. So does the Evolv Technology order (November 2024) over misleading AI weapons-scanner claims. So does Cleo AI’s $17 million settlement (March 2025) for deceptive AI-powered cash-advance marketing.
The pattern: orders tied to consumer harm and dollars are surviving. Orders tied to fuzzy quality claims about the AI itself are getting pulled. Rytr’s underlying complaint was that the product could generate “deceptive testimonials” — a thin harm theory even the agency now seems to regret.
For founders shipping AI features, this is actually worse news than a simple “enforcement is over.” You can’t read the tea leaves. You can’t look at the Rytr reversal and assume your own AI-marketing claims are safe. The FTC is still running Operation AI Comply. It’s just more selective about which cases to keep.
What changes Monday morning: don’t let your marketing team celebrate. If your AI claims involve a specific dollar outcome for a consumer — “save $X,” “earn $Y,” “get approved for $Z” — the FTC is still watching. If your claims are about model quality or capability without a consumer harm hook, you have slightly more room. Slightly.
— Nathan Zakhary