On July 1, 2026, the Federal Trade Commission published a proposed Policy Statement on AI Accuracy and Output Steering that every AI founder in America needs to read before the July 31 public comment deadline. The core message is blunt: if your AI system steers outputs toward undisclosed ideological, political, or commercial objectives, you may already be violating Section 5 of the FTC Act. The commission backed its position with a striking statistic: consumers accept AI outputs without independent fact-checking more than 90% of the time, according to the FTC's own data. That level of trust, the agency argues, creates a reasonable expectation that AI systems prioritize accuracy over hidden agendas. The proposed statement is the most significant federal regulatory action on AI output transparency since the Biden-era Executive Order, and it positions the FTC as the primary AI regulator through existing consumer protection authority rather than new legislation.
What the FTC Policy Statement Actually Says
The proposed statement was issued under Executive Order 14365, signed by President Trump on December 11, 2025, which directed the FTC to clarify how Section 5 of the FTC Act applies to AI models. The FTC's position rests on its three-part deception framework: conduct is deceptive if it involves a representation, omission, or practice that is likely to mislead a reasonable consumer in a material way. The commission argues that AI companies have made both explicit and implicit representations that their systems are designed to produce the best, most accurate, and most faithful output possible within their technological and resource constraints. When a company trains or configures its model to pursue undisclosed ideological objectives, the FTC concludes that practice may be deceiving consumers regardless of the motivation, whether the company acts on its own volition or in response to a state law requirement.
A critical distinction in the statement separates intentional ideological steering from AI hallucinations. Hallucinations arise from technological and resource limitations rather than design decisions and do not, by themselves, raise Section 5 concerns. The FTC is not penalizing imperfect AI. It is targeting intentional design choices that suppress or distort outputs in undisclosed ways. That distinction matters enormously for founders building in this space: technical failures are one thing, but intentional steering without disclosure is a regulatory landmine.
The State Law Preemption Gambit
One of the most significant features of the proposed statement is its treatment of state AI laws. The FTC singled out Colorado's Artificial Intelligence Act as an example of a state law that may pressure AI companies to suppress output accuracy to avoid disparate impact liability. The statement concludes that such a state law is impliedly preempted to the extent it conflicts with the federal regulatory scheme established by Section 5 of the FTC Act. The argument is straightforward: a state law that effectively requires an AI company to deceive its consumers cannot stand alongside federal law designed to protect consumers from deception. This creates a direct tension that courts will likely have to resolve. For AI founders, this means compliance planning just got more complicated. If you operate in Colorado or other states with their own AI laws, you may face conflicting requirements between state nondiscrimination obligations and federal truthfulness standards. The FTC is signaling that federal law wins, but the legal uncertainty will persist until courts rule.
The Disclosure Safe Harbor and Why It Is Harder Than You Think
The proposed statement acknowledges that AI companies can avoid Section 5 liability through clear, conspicuous, and adequate disclosures. If you disclose that your system is designed to prioritize certain objectives over what users would otherwise expect, the FTC says that can shift reasonable consumer expectations and insulate you from deception claims. But here is the catch: the FTC sets an extremely high bar for what counts as adequate disclosure. A disclaimer buried in terms of service will not suffice. The disclosure must be sufficiently prominent and persistent to actually shift the reasonable consumer expectations created by the company's broader marketing and the inherent nature of AI as a problem-solving tool. The FTC explicitly notes that the more a disclosure cuts against consumers' reasonable expectations established in other contexts, the more prominent and persistent that disclosure must be. In plain English: if you are training a customer service chatbot to prioritize upselling over accuracy, you need to tell users upfront in a way they cannot miss, and one line buried in a privacy policy will not cut it.
Three Things Founders Need to Do Before the Comment Deadline
The public comment period closes July 31, 2026, which means founders have a narrow window to shape the final rule. Here is the actionable checklist. First, audit your AI system's output steering mechanisms. Document every place where your model or application intentionally prioritizes one outcome over accuracy, whether for commercial goals, safety constraints, or ideological alignment. If you cannot explain why your AI said what it said, you have a disclosure problem. Second, review your existing disclosure practices against the FTC's high bar. Where are your disclosures placed? Are they prominent enough to actually change user expectations? If your answer involves the phrase It is in our terms of service, you need a redesign. Third, submit a public comment before July 31. The FTC is actively seeking input from industry participants, and early engagement shapes the final language. The proposed statement is just that, a proposal, and the commission has shown willingness to adjust based on feedback. But silence will be interpreted as consent.
For founders building consumer-facing AI products, this policy statement is the clearest signal yet that the regulatory landscape is shifting from voluntary guidelines to enforceable rules. The FTC is using existing consumer protection law to fill the gap left by stalled federal AI legislation, and the accuracy disclosure requirement is likely to be the first wave of enforceable AI-specific regulation. The July 31 comment deadline is not just a procedural milestone. It is the last chance to help define what compliance looks like before the rules become final.