FTC Seeks Comment on AI Accuracy Policy

AI outputs could be shaped to mislead, the FTC says.

The Federal Trade Commission is inviting public input on a proposed policy statement that would treat distortions of AI system outputs to advance undisclosed ideological objectives as possible unfair or deceptive conduct under the FTC Act. The Commission frames the issue around how AI companies may manipulate the behavior of their models in ways that run counter to consumer expectations for objectivity and accuracy. The policy statement argues that such distortions could violate Section 5 of the FTC Act by misrepresenting the effectiveness and suitability of AI systems for different tasks.

The filing makes clear that the agency intends to hear from both businesses and consumers about experiences with AI systems that have been subverted to serve ideological aims. “The FTC wants to hear from businesses and consumers about their experiences and concerns regarding the subversion of AI systems for ideological ends,” said FTC Chair Andrew N. Ferguson. He added that the input will help the Commission craft a final policy. The Commission’s stated goal, the filing notes, is to advance what President Donald Trump has described as a push to expand America’s global dominance in artificial intelligence.

A key feature of the policy statement is how it treats state-level efforts to regulate AI. The Commission contends that state laws that demand altering the truthful outputs of AI models may be impliedly preempted if they conflict with a federal regulatory scheme. The point matters in a landscape where some states have moved to impose ideological objectives through AI rules. The document also links the inquiry to a December executive order directing the FTC to issue a policy statement on the legal implications of such state laws. In short, the FTC is signaling that federal rules may supersede state attempts to bend AI outputs, at least where those attempts conflict with federal safeguards for truthful results.

For compliance teams and AI product leaders, the move signals a sharpened enforcement lens on how AI systems are presented to users. If the policy statement becomes final, firms could face scrutiny not only for explicit misrepresentations about capabilities but also for subtler shifts in outputs that steer consumer behavior toward undisclosed goals. That creates incentives to tighten governance around model outputs, disclosures, and testing for objectivity. Companies will want robust documentation showing how model decisions are aligned with stated representations and how any automated changes to outputs are justified, disclosed, and auditable.

Industry observers should watch two levers closely: the scope of what counts as a deceptive output and how strictly the FTC will tie that to consumer representations. The policy’s emphasis on ideological objectives implies that motive and disclosure will matter in enforcement, not just accuracy in a narrow sense. In practice, this means risk assessments for AI deployments may need to account for potential perception risks, how users interpret recommended results, predictions, or automated conclusions as to reliability and bias. It also means board and governance measures may require clearer signoffs on what outputs a product team claims to deliver and under what conditions any alterations are permissible.

As the comment period unfolds, practitioners should prepare for a long arc of clarification and guardrails. The final policy statement could set a framework that shapes testing protocols, marketing claims, and compliance audits across AI-enabled products. The FTC’s approach will likely influence how companies balance speed and experimentation with the obligation to avoid hidden manipulation that could mislead consumers about AI effectiveness or honesty.

Sources