Preventing AI Censorship Act

This Act may be cited as the “Preventing AI Censorship Act”.

(a) In general.— A Federal employee who, under color of any statute, ordinance, regulation, custom, or usage, of the United States, engages in covered conduct thereby subjecting, or causing to be subjected, any citizen of the United States to the deprivation of any rights, privileges, or immunities secured by the First Amendment, shall be liable to the party injured in an action at law, suit in equity, or other proper proceeding for redress.

(b) Exception.— This Act does not authorize a Federal employee to bring a suit against their Federal employer or the Federal Government for conduct that is within the scope of the employment relationship. Nothing in this Act shall be construed to limit any otherwise lawful law enforcement activity conducted pursuant to a valid warrant, court order, or other judicial authorization.

(c) Attorney’s fees.— In any action or proceeding to enforce this Act, the court, in its discretion, may allow the prevailing party, other than the United States, a reasonable attorney’s fee as part of the costs.

(d) Rule of construction.— Nothing in this Act may be construed to limit the ability of an injured party to seek damages, injunctive relief, or any other form of redress for the deprivation of any rights, privileges, or immunities secured by the First Amendment, or by any other provision of the Constitution.

(e) Definitions.— In this section:

(1) The term Federal employee means an individual, other than the President or the Vice President, who occupies a position in any agency or instrumentality of the executive branch (including any independent agency).

(2) The term artificial intelligence has the meaning as outlined in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).

(3) The term artificial intelligence provider means any person or entity that develops, deploys, operates, hosts, or maintains artificial intelligence, whether for compensation or without charge.

(4) The term covered conduct means—

(A) any action by a Federal employee to coerce, compel, direct, induce, or encourage an artificial intelligence provider to:

(i) ban, suppress, remove, alter, or otherwise restrict content, outputs, or responses generated by artificial intelligence based on viewpoint, partisan affiliation, religious belief or practice, ideology, perceived truth or falsity, or perceived bias;

(ii) modify the training data, model weights, fine-tuning procedures, system prompts, safety classifiers, or operational parameters of artificial intelligence for the purpose of filtering, distorting, or suppressing expression based on viewpoint, partisan affiliation, religious belief or practice, ideology, perceived truth or falsity, or perceived bias;

(iii) deny, degrade, restrict, or alter access to artificial intelligence, or provide a materially inferior quality of service through artificial intelligence, to any person based on that person’s viewpoint, political affiliation, religious belief or practice, ideology, perceived truth or falsity, or perceived bias; or

(iv) collect, retain, report, or disclose information about a user’s prompts, queries, or interactions with artificial intelligence to any Federal agency or employee, where such collection, retention, or disclosure is undertaken for the purpose of, or with the reasonably foreseeable effect of, surveilling, chilling, or penalizing that user’s lawful expression; or

(B) any action by a Federal employee to directly interfere with an individual’s lawful use of artificial intelligence.

(f) Severability.— If any provision of this Act or the application of a provision of this Act to any person or circumstance is held to be unconstitutional, the remainder of this Act, and the application of the provisions to any person or circumstance, shall not be affected thereby.