FTC TELLS AI COMPANIES THAT SECRETLY STEERING RESPONSES AWAY FROM THE TRUTH IS ILLEGAL AND WANTS THE PUBLIC TO WEIGH IN The Federal Trade Commission published a proposed policy statement on July 1 that could reshape how every major AI company designs its products, and it has given the public until July 31 to respond. The statement is pointed. If an AI company is secretly steering its model to give answers that serve the company’s interests rather than the user’s actual question, the FTC believes that is deception under federal law. The concern is not hypothetical. The FTC flagged that consumers accept AI output without independent fact-checking more than 90 percent of the time. They trust that the system is giving them the best answer it can. If the system is instead quietly optimizing for something else, whether that is avoiding liability, protecting advertiser relationships, or advancing a political position the company prefers, users have no way to know. The proposed statement invokes Section 5 of the FTC Act, which prohibits unfair or deceptive trade practices. It would apply to any company that has made explicit or implied representations that its AI is designed to give accurate, faithful responses. In practice, that is every major AI company operating today. The statement also takes aim at Colorado’s AI Act, arguing that state laws pushing companies to adjust AI outputs for equity reasons may conflict with federal consumer protection requirements. The FTC is essentially telling both AI companies and state legislatures that accuracy is not negotiable. This is the sharpest federal action yet aimed at what actually happens inside an AI model, not just how it is marketed on the outside. Keywords: FTC AI accuracy, AI deception federal law, AI output steering, Section 5 FTC Act