FTC Proposes Policy Statement Targeting AI Output Accuracy and "Ideological Manipulation"
Following the close of the public comment period on July 31, 2026, the Federal Trade Commission (FTC) is facing a massive, highly unusual, and politically diverse backlash over its proposed Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems (File No. P264200, Docket FTC-2026-0859). Published on July 7, 2026, the proposed statement warns that AI developers who quietly steer or manipulate AI model outputs away from user expectations—even to comply with state laws or mitigate bias—may be committing deceptive trade practices in violation of Section 5 of the FTC Act.
The Core of the Proposed Policy Statement
The FTC's proposed policy statement distinguishes between technological limitations (such as accidental "hallucinations" or errors) and deliberate design choices that suppress accuracy or redirect outputs toward an undisclosed objective. Under the proposed policy, companies that quietly modify AI outputs to advance ideological goals, political agendas, or other undisclosed objectives face potential Section 5 liability.
The FTC applies its standard three-part deception test to these practices:
- Likely to Mislead: The FTC argues that there is an implicit representation that AI systems aim to produce the most accurate and objective output possible. Quietly steering outputs away from this expectation is a material omission or representation.
- Reasonable Consumer Perspective: The FTC notes that more than 90% of users accept AI outputs without independent fact-checking, making their reliance on the model's accuracy highly reasonable.
- Materiality: The accuracy and objectivity of the AI's output are material to the consumer's decision to use or pay for the service.
The Clash with State AI Laws (e.g., Colorado's AI Act)
The most controversial and legally complex aspect of the proposed policy statement is its direct target on state-level AI regulations.
The FTC explicitly declared that AI developers who steer system outputs "contrary to consumers’ reasonable expectations … including attempted compliance with a State law, such as Colorado’s recently revised Artificial Intelligence Act … may deceive consumers in violation of Section 5 of the FTC Act."
This creates an extraordinary regulatory conflict:
- State Mandates: Laws like Colorado’s Artificial Intelligence Act (originally SB 24-205, revised by SB 26-189) require developers and deployers of high-risk AI systems to implement bias-mitigation measures, explain "adverse decisions," and actively steer models to prevent disparate impacts or discriminatory outcomes.
- FTC Stance: Under the proposed policy statement, if a company modifies its AI's outputs to comply with these state-level anti-discrimination or bias-mitigation mandates without explicit, prominent disclosure to the consumer, the FTC may prosecute them for deceptive practices under federal law. The FTC's proposed solution is absolute disclosure regarding how and why a model's outputs are being steered.
Industry Backlash and Legal Challenges
During the public comment period, which closed on July 31, 2026, a wide range of industry groups, legal foundations, and policy institutes filed strenuous objections, setting the stage for immediate federal litigation if the FTC votes to finalize the statement.
- Washington Legal Foundation (WLF): Filed a formal comment urging the FTC to withdraw the "unconstitutional" proposal. WLF argued that the FTC is attempting to regulate protected speech and "ideological bias" under the guise of consumer protection, exceeding its statutory authority under Section 5.
- Abundance Institute: Argued that the FTC’s proposed policy statement is overbroad, fails to distinguish between standard safety/compliance guardrails and deceptive practices, and would chill the development of safe AI systems by exposing developers to arbitrary enforcement.
- Compliance and Operational Risks: Critics point out that the FTC's policy could make it nearly impossible for developers to balance competing legal demands—forcing them to choose between state-mandated bias mitigation and federal FTC deception charges.