How Is This Bill Enforced
Verbatim statutory text on the left; plain-language analysis and a per-section checklist on the right. Numbered markers cross-link to the matching checklist row.
This Act may be cited as the "Reliable Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) Research Act of 2025".
Establishes the short title of the Act as the Reliable Artificial Intelligence Research Act of 2025. No compliance obligations are created.
(1) ADVERSARIAL ROBUSTNESSAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1).—The term "adversarial robustnessAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1)" means the degree to which an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.
(2) ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2).—The term "artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2)" has the meaning given the term in section 5002 of the National Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) Initiative Act of 2020 (15 U.S.C. 9401).
(3) INTERPRETABILITYInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3).—The term "interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3)" means the degree to which humans are able to accurately understand how an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.
(4) RED-TEAMINGRed-teamingThe term "red-teaming" means a structured, interactive, and adversarial process to test an artificial intelligence system by simulating real-world actions to find vulnerabilities or flaws in the system.Sec. 2(4).—The term "red-teamingRed-teamingThe term "red-teaming" means a structured, interactive, and adversarial process to test an artificial intelligence system by simulating real-world actions to find vulnerabilities or flaws in the system.Sec. 2(4)" means a structured, interactive, and adversarial process to test an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) system by simulating real-world actions to find vulnerabilities or flaws in the system.
(5) SECRETARYSecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5).—The term "SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5)" means the Secretary of Homeland Security.
Defines five terms used throughout the Act: adversarial robustness, artificial intelligence (by cross-reference to the National AI Initiative Act of 2020), interpretability, red-teaming, and Secretary. These definitions set the technical scope of the prize competitions but impose no compliance obligations on private parties.
(a) 1 PRIZE COMPETITION REQUIRED.—Not later than 270 days after the date of enactment of this Act, the Secretary of Homeland Security shall commence carrying out at least 1 prize competition under section 24 of the Stevenson-Wydler Technology Innovation Act of 1980 (15 U.S.C. 3719) to advance the science of interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3) in a manner relevant to commercially available or widely used artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) products.
(b) 1 CONSULTATION.—In carrying out the prize competition required by subsection (a), the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall consult with— (1) the Secretary of Commerce; (2) the Director of the National Institute of Standards and Technology; (3) the National Cyber Director; (4) the Director of the National Science Foundation; and (5) any industry expert from the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) sector in the United States that the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) considers relevant.
(c)(1)–(3) 1 IN GENERAL.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall develop the structure and evaluation criteria for a prize competition carried out under subsection (a) in accordance with the primary purpose described in that subsection. (2) COMPETITION STRUCTURE.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) may— (A) structure a competition under subsection (a) into 1 or more phases, including submission of interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3) frameworks, submission of interpretable artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) models, and unique basic research; and (B) open these phases to the same, or to distinct, contestant pools. (3) EVALUATION CONSIDERATIONS.—In developing the evaluation criteria for the frameworks, models, or methods submitted for a prize competition under subsection (a), the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall consider— (A) the degree to which a submission advances broadly applicable principles of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3); (B) the practical value of a submission in making artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) more understandable and reliable in high-risk, high-value use cases; and (C) the likelihood that the unique research submitted will create standards for artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3) in the government or industry.
(d) PROGRAM ADMINISTRATION.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) may enter into contracts, cooperative agreements, or other agreements with for-profit or nonprofit entities or State, territorial, local, or Tribal agencies to design and administer any prize competition carried out under subsection (a).
Requires the Secretary of Homeland Security to commence at least one prize competition within 270 days of enactment to advance the science of AI interpretability as applied to commercially available or widely used AI products. The competition must be carried out under the Stevenson-Wydler Technology Innovation Act's prize authority (15 U.S.C. 3719). The Secretary must consult with the Secretary of Commerce, the Director of NIST, the National Cyber Director, the Director of the National Science Foundation, and relevant industry experts.
The Secretary has discretion to structure the competition in phases — including interpretability framework submissions, interpretable AI model submissions, and basic research — and to open phases to different contestant pools. Evaluation criteria must consider how well submissions advance broadly applicable interpretability principles, their practical value for high-risk use cases, and the likelihood of creating government or industry standards. The Secretary may contract with external entities to administer the competition.
(a) 2 PRIZE COMPETITION REQUIRED.—Not later than 270 days after the date of enactment of this Act, the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall commence carrying out at least 1 prize competition under section 24 of the Stevenson-Wydler Technology Innovation Act of 1980 (15 U.S.C. 3719) to develop capable artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) models that are designed to exhibit adversarial robustnessAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1) in circumstances necessary for at least 1 high-impact, high-risk application in government or industry.
(b) 2 CONSULTATION.—In carrying out a prize competition required by subsection (a), the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall consult with— (1) the Secretary of Commerce; (2) the Director of the Institute of Standards and Technology; (3) the National Cyber Director; (4) the Director of the National Science Foundation; (5) any industry expert from the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) sector in the United States that the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) considers relevant; and (6) the head of any Federal agency who has authority or expertise in a high-impact, high-risk application of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) that could be an appropriate subject for a prize competition under subsection (a).
(c)(1)–(3) 2 IN GENERAL.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall develop the structure and evaluation criteria for a prize competition carried out under subsection (a) in accordance with the primary purpose described in that subsection. (2) COMPETITION STRUCTURE.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) may— (A) structure a competition under subsection (a) into 1 or more phases, including submission of adversarial robustnessAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1) frameworks, submission of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) models, and red-teamingRed-teamingThe term "red-teaming" means a structured, interactive, and adversarial process to test an artificial intelligence system by simulating real-world actions to find vulnerabilities or flaws in the system.Sec. 2(4); and (B) open these phases to the same, or to distinct, contestant pools. (3) EVALUATION CONSIDERATIONS.—In developing the evaluation criteria for the frameworks, models, or methods submitted for a prize competition under subsection (a), the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall consider— (A) the degree to which a submission advances broadly applicable principles of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2) robustness; and (B) the practical value of the submission in reducing the risk of adversarial attacks in high-risk, high-value use cases of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" has the meaning given the term in section 5002 of the National Artificial Intelligence Initiative Act of 2020 (15 U.S.C. 9401).Sec. 2(2).
(d) PROGRAM ADMINISTRATION.—The SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) may enter into contracts, cooperative agreements, or other agreements with for-profit or nonprofit entities or State, territorial, local, or Tribal agencies to design and administer any prize competition carried out under subsection (a).
Mirrors the structure of Section 3 but targets adversarial robustness rather than interpretability. The Secretary must commence at least one prize competition within 270 days to develop AI models that exhibit adversarial robustness for at least one high-impact, high-risk government or industry application. The consultation requirement adds the head of any federal agency with authority or expertise in high-impact AI applications. Evaluation criteria focus on advancing broadly applicable robustness principles and the practical value of reducing adversarial attack risk in high-risk AI use cases.
(a) 3 IN GENERAL.—Not later than 180 days after the date on which the first prize competition concludes, the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) shall submit to the appropriate congressional committees a report that includes— (1) an evaluation of how the results of the competitions inform the fields of interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3) and adversarial robustnessAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1); (2) an assessment of any gaps in these fields identified by the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) over the course of the competitions; and (3) any suggested action that Congress should take to advance the fields of interpretabilityInterpretabilityThe term "interpretability" means the degree to which humans are able to accurately understand how an artificial intelligence model makes decisions and considers inputs and how the outputs or behaviors of the model respond to a change in the inputs.Sec. 2(3), adversarial robustnessAdversarial robustnessThe term "adversarial robustness" means the degree to which an artificial intelligence model is able to resist attacks that would induce it to produce incorrect, restricted, or harmful outputs, while maintaining integrity, reliability, and privacy.Sec. 2(1), and any related research.
(b) APPROPRIATE CONGRESSIONAL COMMITTEES DEFINED.—In this section, the term "appropriate congressional committees" means— (1) the Committee on Homeland Security and Governmental Affairs of the Senate; and (2) the Committee on Homeland Security of the House of Representatives.
Requires the Secretary to submit a report to the Senate Committee on Homeland Security and Governmental Affairs and the House Committee on Homeland Security within 180 days after the first prize competition concludes. The report must evaluate how the competition results inform interpretability and adversarial robustness, assess gaps identified during the competitions, and suggest legislative action Congress should take to advance these fields.
There is authorized to be appropriated to the SecretarySecretaryThe term "Secretary" means the Secretary of Homeland Security.Sec. 2(5) to carry out this section $10,000,000 for the period of fiscal years 2026 through 2030.
Authorizes $10,000,000 in appropriations for fiscal years 2026 through 2030 for the Secretary to carry out the Act. This is an authorization of appropriations, not a compliance obligation.