FAIR Act

This bill would require each federal agency to buy only large language models (a type of AI that writes text in response to prompts) that are developed according to a set of principles. The principles say the models must be truthful when asked for facts, prioritize historical accuracy, scientific inquiry, and objectivity, and acknowledge uncertainty when information is incomplete. The models must also be neutral and nonpartisan and not shape responses in favor of ideologies such as diversity, equity, and inclusion. Developers could not intentionally build partisan or ideological judgments into model output unless the user prompts for them or can readily access them. The rules cover executive departments, military departments, independent agencies, and wholly owned government corporations, but not the Government Accountability Office.

Full text and official record on Congress.gov

Sponsors

• Harriet M. Hageman Republican sponsor
• 1 cosponsors