FTC Proposes AI Truthfulness Rules for Deceptive Output

The FTC just proposed rules to curb deceptive AI practices.

The filing states that the Federal Trade Commission is proposing a policy statement to apply the prohibition on deceptive acts or practices in Section 5 of the FTC Act to companies that market artificial intelligence systems. In plain terms, the agency wants to treat misrepresentations about what AI can or will do as a potential violation of the law, not just a bad marketing tactic. The policy is titled the Policy Statement Concerning the Suppression of Accuracy in Artificial Intelligence Systems, signaling a focus on how some firms may alter, withhold, or spin AI outputs to fit business aims or regulatory pressures. The notice foregrounds a core concern: steering AI systems away from what users reasonably expect, and the risk that customers will be harmed by inaccurate or misleading results.

For compliance teams, the shift could mean new guardrails around claims about accuracy, reliability, and performance. The filing states that AI marketers would be held to the same deception standard as any other product manufacturer or service provider. That could require more robust disclosures about limits and known gaps in AI outputs, independent verification of performance claims, and more stringent documentation of how models are tested before release. The policy’s emphasis on accuracy suggests a higher bar for risk disclosures, especially where outputs could influence financial decisions, health and safety, or consumer financial products. In practice, this would push companies to implement stronger governance around data provenance, model versioning, and audit trails that demonstrate why a claim about an AI system is credible.

The policy also notes that the issue is not limited to a single product but to the broader ecosystem of AI offerings marketed to consumers and businesses. The White House’s framing of AI and the divergence in economic outcomes, along with state level actions like Colorado’s revised Artificial Intelligence Act, illustrate a regulatory landscape that is increasingly active on two fronts: robust truthfulness in marketing and safeguards against output manipulation to meet external pressures. Colorado’s act is cited as part of the broader context in which firms may face competing or misaligned incentives to steer outputs to satisfy laws or risk preferences, potentially triggering the FTC to step in if consumer expectations are misled.

Industry observers should watch for two practical implications. First, enforcement mechanics will matter as much as any rule text. The filing centers on deception under Section 5, which means the FTC could pursue investigations, settlements, or injunctive remedies when a company makes false claims about an AI system or uses steering tactics that degrade accuracy. Second, deadlines and the shape of any compliance timetable remain to be set in forthcoming guidance. The notice signals that the policy would guide how marketers must align product claims with actual capabilities, but exact timelines for implementation and what qualifies as a breach are not spelled out in the filing itself.

From a practitioner standpoint, there are concrete takeaways. The first is governance: build auditable processes for testing AI outputs, logging data sources, and validating claims with independent verifications before marketing. The second is disclosure discipline: craft transparent communications that clearly delineate what the AI can and cannot do, including known limitations and the conditions under which outputs may vary. The third is supply chain vigilance: scrutinize third party components and data partners for their own accuracy assurances, because misrepresentations can ripple through a product that relies on external inputs. The fourth is readiness for evolving scrutiny: establish an internal review cadence that anticipates FTC questions about accuracy, outputs, and user expectations, given the rising regulatory focus on AI honesty.

As regulators finalize the policy, compliance leaders should align product roadmaps with a more explicit truthfulness standard, balancing user trust, product performance, and the legal risk of deception claims.