FTC’s Proposal Warns Businesses On Accuracy Of AI Output

The Federal Trade Commission (FTC or the Commission) released a policy statement concerning the suppression of accuracy in artificial intelligence (AI) systems on July 1, 2026 (Policy Statement), addressing how existing consumer protection laws under Section 5 of the FTC Act apply to accurate output from AI systems. Specifically, the Policy Statement explains that AI systems that manipulate or suppress output accuracy without transparent disclosure face possible violations of laws for deceptive or unfair trade practices under existing law.

The Policy Statement was issued pursuant to Executive Order 14365, “Ensuring a National Policy Framework for Artificial Intelligence,” which directed the Commission to clarify how Section 5 of the FTC Act applies to AI models and how it interacts with state laws that require AI companies to alter model outputs.

AI companies market their products as tools designed to produce the most accurate output possible, which consumers are then reasonably relying on without questioning or independently fact-checking those outputs. If a business trains or configures an AI model to deprioritize accuracy in favor of an undisclosed objective, including an ideological or equity objective adopted to satisfy a state law, the FTC views that as a material, deceptive omission actionable under Section 5.

The Preemption Argument

As an example of a state regime that could pressure companies into altering outputs to manage disparate impact exposure, the Policy Statement singled out Colorado’s revised Artificial Intelligence Act, which allows liability for discriminatory outcomes caused by a customer’s use of an AI product. The FTC’s position was clear – compliance with a state law is not a defense to a Section 5 deception claim, and the FTC believes that such state mandates are impliedly preempted to the extent they conflict with Section 5’s prohibition on deceiving consumers. A company vacillating between a state accuracy-altering mandate and the federal deception liability is, in the FTC’s view, still bound to either disclose the alteration or avoid it.

Liability Triggers

The liability trigger is not only limited to AI developers. Based on the broad prohibition against deceptive acts and practices, liability would extend to any company that promotes, licenses, or makes representations about the accuracy of an AI system. According to the Policy Statement, “a company could be tempted, for example, to abuse consumer trust by training a model surreptitiously to produce ideologically motivated distortions in a response to a factual question.” Accordingly, the following actions could trigger liability:

• Training or steering an AI model to produce outputs that deprioritize accuracy in favor of an undisclosed ideological, political, or equity objective.
• Altering outputs to manage state law disparate impact exposure without disclosing that the alteration affects accuracy.
• Marketing a system as accurate, neutral, or the best possible answer while quietly optimizing for a different, undisclosed objective.

Ordinary hallucinations or errors stemming from technical/resource limitations are outside the scope of the Policy Statement. 

The Disclosure Safe Harbor

The Policy Statement allows companies to shape consumer expectations through disclosure, but sets the bar high based on existing consumer deception case law. Company disclosures must be clear, conspicuous, persistent, and prominent, rather than buried in the terms of service or a one-time notice hidden in fine print.

Next Steps for Businesses to Consider

While this is a proposed Policy Statement, and is still subject to a comment period through July 31, 2026, there are several steps companies should take now to get in front of this regulatory enforcement signal:

• Inventory the current output-steering, fine-tuning, and safety layer practices you or your vendors employ, and map each to a business or safety justification.
• Evaluate whether any steering is attributable to state-law compliance and, if so, assess disclosure obligations and preemption exposure with counsel.
• Review consumer-facing marketing and disclosures for consistency with actual model behavior.
• Assess how this regulatory expectation could impact your marketing or advertising strategy, if an AI system promotes your product or services based on platform sponsorship instead of accuracy.
• Monitor for a final Policy Statement and any related enforcement activity or state legislative/regulatory response.

The FTC’s Policy Statement signals a move toward stricter oversight of AI practices, emphasizing transparency and fairness. Businesses that prioritize transparency, fairness, and compliance with existing laws will be better positioned to navigate this evolving landscape and mitigate legal risks.

To read the full Policy Statement, click here.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.