US FTC targets companies that secretly cripple AI accuracy in new policy move

On August 27, the US Federal Trade Commission (FTC) released a proposed policy statement warning AI developers that deliberately degrading the accuracy of their systems could violate consumer protection and competition laws. The agency is targeting scenarios where companies quietly weaken or hide certain model capabilities—sometimes to dodge regulatory scrutiny or public criticism—without being transparent with users or regulators about what they have changed and why.

The proposal makes clear that it covers a wide range of AI systems, including generative models used for text and image creation. It also stresses that not only model providers but also downstream service operators who embed these models may fall under the FTC’s watch. Even when companies claim they are reducing capabilities in the name of “safety” or “misinformation control,” the FTC suggests they must still explain the purpose and impact of those changes in a way that is not deceptive or misleading.

The commission will now seek public comments before deciding whether to adopt the statement as formal enforcement guidance. As the White House and Congress debate broader AI legislation, the FTC’s move signals that US regulators are beginning to define concrete accountability standards for how AI systems are tuned, deployed, and messaged to the public.