FAIR Act

Summary

Prohibits Federal agencies from procuring large language models unless developed according to specific unbiased AI principles.

Requires large language models to respond truthfully to factual inquiries.

Mandates these models prioritize historical accuracy, scientific inquiry, and objectivity, acknowledging uncertainties when information is incomplete or contradictory.

Ensures these models remain neutral and avoid manipulating responses towards ideological dogmas like diversity, equity, and inclusion.

Prohibits developers from intentionally encoding partisan or ideological judgments into model outputs unless prompted by or accessible to the end user.

Defines "agency" to include executive departments, military departments, independent establishments, and wholly owned government corporations, excluding the Government Accountability Office.

Defines "large language model" as AI models trained on diverse datasets to produce natural-language responses to prompts.

Prohibits federal procurement of large language models not developed with unbiased AI principles.

A BILL To prohibit the Federal procurement of large language models not developed in accordance with unbiased AI principles, and for other purposes.

Be it enacted by the Senate and House of Representatives of the United States of America in Congress assembled,

SECTION 1. Short title.

This Act may be cited as the “Fair Artificial Intelligence Realization Act of 2025” or the “FAIR Act”.

Requires agency heads to procure only large language models adhering to truthfulness, accuracy, neutrality, and transparency principles.

SEC. 2. Prohibition on Federal procurement of certain large language models.

(a) Policy of the United States.—It is the policy of the United States to promote the innovation and use of trustworthy artificial intelligence.

(b) In general.—To advance the policy established under subsection (a), the head of each agency shall, procure after the date of the enactment of this Act only those large language models developed in accordance with the following principles:

(1) Large language models shall be truthful in responding to user prompts seeking factual information or analysis.

(2) Large language models shall prioritize historical accuracy, scientific inquiry, and objectivity, and shall acknowledge uncertainty where reliable information is incomplete or contradictory.

(3) Large language models shall be neutral, nonpartisan tools that do not manipulate responses in favor of ideological dogmas such as diversity, equity, and inclusion.

(4) Developers of large language models shall not intentionally encode partisan or ideological judgments into a large language model output unless those judgments are prompted by or otherwise readily accessible to the end user.

Defines "agency" and "large language model" related to AI legislative context.

(c) Definitions.—In this Act:

(1) AGENCY.—The term “agency”—

(A) means—

(i) an executive department (as such term is defined in section 101 of title 5, United States Code);

(ii) a military department (as such term is defined in section 102 of title 5, United States Code);

(iii) an independent establishment (as such term is defined in section 104 of title 5, United States Code); and

(iv) a wholly owned Government corporation (as such term is defined in section 9101 of title 31, United States Code); and

(B) does not include the Government Accountability Office.

(2) LARGE LANGUAGE MODEL.—The term “large language model” means a generative AI model trained on vast, diverse datasets that enable the model to generate natural-language responses to user prompts.