What happened
On 25 September 2026 at the Reuters Momentum AI conference in Austin, FTC Chairman Andrew Ferguson said he will 'continue as long as I am chairman to resist this anthropomorphizing of these tools,' rejecting treatment of AI agents as autonomous actors 'that break loose' with 'wills and desires of their own.' He asserted that where an AI tool carries out the instructions it was given, the developer is the liable party, and that existing legal tools — including FTC authority over failure to disclose data breaches — can reach AI developers. He also announced the FTC is preparing to request data from consumer-facing companies (delivery apps, rideshare, airlines) for a market study on personalized pricing.
Why it matters
This is the clearest federal enforcement-direction signal yet on agentic AI liability. Consumer agents now on the market (Meta's Muse, xAI's GrokBot, Apple's Siri AI) operate under divergent liability models; Ferguson's position means the FTC will treat developer instruction and audit-trail evidence as the basis for enforcement, and the 'rogue agent' defense will not be entertained. It also sharpens the agency's trajectory following its proposed enforcement policy statement on AI-driven pricing discrimination.
Action needed
AI developers shipping agents should prepare audit-trail evidence tying agent actions to developer instructions, review agent terms against FTC unfair/deceptive-practice exposure, and treat breach-notification/Safeguards Rule obligations as applicable to AI systems holding personal data.