FIRE statement on suppression of accuracy in artificial intelligence systems

The following statement can be attributed to John Coleman, FIRE’s legislative counsel for AI and free expression.

The FTC is considering a policy that would expand how it regulates AI systems.

Section 5 of the FTC Act targets unfair business practices, but this proposal threatens to give the government power to control AI-generated expression that risks violating the First Amendment. AI developers shape their systems through choices about training data, system prompts, alignment, and other design decisions — editorial judgments protected by the First Amendment. The FTC’s proposal assumes there is one “best output” and one set of user expectations. 

The FTC says it wants to prevent AI systems from being ideologically “steered.” But determining whether an AI has been improperly steered would require the government to judge how information is selected, organized, and presented on contested issues in areas like public health, politics, and history. This approach would pressure developers to suppress lawful speech, blur protected editorial judgment, and stretch Section 5 to police AI content. It was designed to target deceptive commercial practices, not the rights of AI developers and users. 

FIRE urges the FTC to reject this interpretation.

See our full filing below.