FTC Floats AI Policy Aiming To Ensure That AI Makers Disclose The Truth About Biases In Their LLMs - Forbes

The proposal is grounded in the FTC Act's consumer protection mandate, which prohibits unfair or deceptive practices. The policy targets the gap between what AI systems claim to do and what they actually produce, particularly when biased training data or model design leads to inaccurate or misleading statements.

The FTC's proposal would compel companies to identify and report specific biases in their LLMs that contribute to factual inaccuracies or harmful stereotypes. This includes biases in training data, algorithmic weighting, and prompt handling that systematically skew outputs. The agency argues that without such disclosures, consumers and businesses cannot make informed decisions about using AI tools. The proposal builds on existing FTC guidance on AI and algorithmic transparency, extending it to explicitly cover truthfulness failures.

The policy is now open for public comment before any final rulemaking. Industry groups have expressed mixed reactions, with some arguing that the disclosure requirements are too vague or technically infeasible given the opaque nature of large models. Consumer advocates generally welcomed the move as a necessary step to hold AI makers accountable. The FTC has not specified a timeline for finalizing the rule.