Federal · Senate Bill · 117th Congress, 1st Session
S1705
S. 1705 — Artificial Intelligence Capabilities and Transparency Act of 2021

Status ● Failed Effective N/A Passage Likelihood N/A

WHAT THIS BILL REGULATES · 2 REQUIREMENT TYPES

How Is This Bill Enforced

Enforcement Authority
No private enforcement mechanism. The bill directs federal agencies (Department of Defense, Director of National Intelligence, Department of Energy, NIST) to carry out specified programs and duties. Oversight is through congressional reporting requirements.
Private Right of Action
No private right of action. Enforcement is exclusive to the designated authority.
Penalties
No damages, penalties, or private remedies specified. The bill creates government organizational and programmatic mandates enforced through internal executive-branch accountability and congressional oversight.

What This Bill Requires

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.

Statutory Text
Analysis & Obligations
Sec. 1
Short title

This Act may be cited as the ''Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Capabilities and Transparency Act of 2021'' or the ''AICT Act of 2021''.

Establishes the short title of the Act as the "Artificial Intelligence Capabilities and Transparency Act of 2021" or the "AICT Act of 2021." This section creates no compliance obligations.

Sec. 2
Findings; Sense of Congress

(a) FINDINGS.—Congress finds the following: (1) By enacting section 1051 of the John S. McCain National Defense Authorization Act for Fiscal Year 2019 (Public Law 115–232), Congress established the National Security Commission on Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) (NSCAI) to ''consider the methods and means necessary to advance the development of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1), machine learning, and associated technologies by the United States to comprehensively address the national security and defense needs of the United States''. (2) The National Security Commission on Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) was comprised of technology and policy experts with combined expertise in their field. (3) The National Security Commission on Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) released its final report to Congress in March of 2021. (4) This Act is primarily based on the consensus recommendations of the National Security Commission on Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1). (5) The National Science Foundation (NSF) established the National Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) (AI) Research Institutes program in 2020 to invest in long-term research into artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1).

(b) SENSE OF CONGRESS.—It is the sense of Congress that— (1) artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1), machine learning, and associated technologies will play a critical role in the economic and national security interests of the United States and its allies; (2) two additional themes for Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Research Institutes should be established in the areas of artificial intelligence safetyArtificial intelligence safetyThe term "artificial intelligence safety" includes technical efforts to improve artificial intelligence systems in order to reduce adverse and unintentional effects of such systems.Sec. 3(3) and artificial intelligence ethicsArtificial intelligence ethicsThe term "artificial intelligence ethics" includes the quantitative analysis of artificial intelligence systems to address matters relating to the effects of such systems on individuals and society, such as matters of fairness or the potential for discrimination.Sec. 3(2); and (3) the Director of the National Science Foundation should stand up the thematic programs described in paragraph (2) as soon as practicable.

This section contains congressional findings regarding the establishment of the National Security Commission on Artificial Intelligence and its final report, and expresses the sense of Congress that NSF should establish two additional AI Research Institute themes in AI safety and AI ethics. No operative compliance obligations are created.

Sec. 3
Definitions

(1)–(6) In this Act: (1) ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1).—The term ''artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1)'' includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting. (2) ARTIFICIAL INTELLIGENCE ETHICSArtificial intelligence ethicsThe term "artificial intelligence ethics" includes the quantitative analysis of artificial intelligence systems to address matters relating to the effects of such systems on individuals and society, such as matters of fairness or the potential for discrimination.Sec. 3(2).—The term ''artificial intelligence ethicsArtificial intelligence ethicsThe term "artificial intelligence ethics" includes the quantitative analysis of artificial intelligence systems to address matters relating to the effects of such systems on individuals and society, such as matters of fairness or the potential for discrimination.Sec. 3(2)'' includes the quantitative analysis of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) systems to address matters relating to the effects of such systems on individuals and society, such as matters of fairness or the potential for discrimination. (3) ARTIFICIAL INTELLIGENCE SAFETYArtificial intelligence safetyThe term "artificial intelligence safety" includes technical efforts to improve artificial intelligence systems in order to reduce adverse and unintentional effects of such systems.Sec. 3(3).—The term ''artificial intelligence safetyArtificial intelligence safetyThe term "artificial intelligence safety" includes technical efforts to improve artificial intelligence systems in order to reduce adverse and unintentional effects of such systems.Sec. 3(3)'' includes technical efforts to improve artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) systems in order to reduce adverse and unintentional effects of such systems. (4) CONGRESSIONAL DEFENSE COMMITTEESCongressional defense committeesThe term "congressional defense committees" has the meaning given such term in section 101(a) of title 10, United States Code.Sec. 3(4).—The term ''congressional defense committeesCongressional defense committeesThe term "congressional defense committees" has the meaning given such term in section 101(a) of title 10, United States Code.Sec. 3(4)'' has the meaning given such term in section 101(a) of title 10, United States Code. (5) INTELLIGENCE COMMUNITYIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5).—The term ''intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5)'' has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003). (6) UNITED STATES PERSONUnited States personThe term "United States person" does not include a member of the Armed Forces or an employee of the Department of Defense.Sec. 3(6).—The term ''United States personUnited States personThe term "United States person" does not include a member of the Armed Forces or an employee of the Department of Defense.Sec. 3(6)'' does not include a member of the Armed Forces or an employee of the Department of Defense.

Defines key terms used throughout the Act, including a broad, multi-prong definition of artificial intelligence and a notable carve-out in the definition of United States person that excludes members of the Armed Forces and employees of the Department of Defense from the privacy and civil liberties protections established in Section 7.

Sec. 4
Digital talent recruiting officers
Government

(a)(1)–(3) 1 DIGITAL TALENT RECRUITING FOR THE DEPARTMENT OF DEFENSE.— (1) DESIGNATION OF CHIEF DIGITAL RECRUITING OFFICER.—Not later than 270 days after the date of the enactment of this Act, the Secretary of Defense shall designate a chief digital recruiting officer within the office of the Under Secretary of Defense for Personnel and Readiness to oversee a digital recruiting office to carry out the responsibilities set forth in paragraph (2). (2) RESPONSIBILITIES.—The chief digital recruiting officer designated under paragraph (1) shall, in coordination with the Director for the Joint Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Center (JAIC), be responsible for— (A) identifying needs of the Department of Defense for specific types of digital talent; (B) recruiting technologists, in partnership with the military services and Department components, including by attending conferences and career fairs, and actively recruiting on university campuses and from the private sector; (C) integrating Federal scholarship for service programs into Department civilian recruiting; and (D) partnering with human resource teams in the military services and Department components to use direct-hire authorities to accelerate hiring. (3) STAFF AND RESOURCES.—The Secretary of Defense shall ensure that the chief digital recruiting officer designated under paragraph (1) is provided with staff and resources sufficient to maintain an office and to carry out the responsibilities set forth in paragraph (2).

(b)(1)–(3) 2 DIGITAL TALENT RECRUITING FOR THE INTELLIGENCE COMMUNITYIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5).— (1) DESIGNATION OF CHIEF DIGITAL RECRUITING OFFICER.—Not later than 270 days after the date of the enactment of this Act, the Director of National Intelligence shall designate a chief digital recruiting officer to oversee a digital recruiting office to carry out the responsibilities set forth in paragraph (2). (2) RESPONSIBILITIES.—The chief digital recruiting officer designated under paragraph (1) shall be responsible for— (A) identifying needs of the intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5) for specific types of digital talent; (B) recruiting technologists, in partnership with components of the intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5), by attending conferences and career fairs, and actively recruiting on college campuses; (C) integrating Federal scholarship for service programs into intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5) recruiting; (D) offering recruitment and referral bonuses; and (E) partnering with human resource teams in the components of the intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5) to use direct-hire authorities to accelerate hiring. (3) STAFF AND RESOURCES.—The Director of National Intelligence shall ensure that the chief digital recruiting officer designated under paragraph (1) is provided with staff and resources sufficient to maintain an office and to carry out the responsibilities set forth in paragraph (2).

(c)(1)–(3) 3 DIGITAL TALENT RECRUITING FOR THE DEPARTMENT OF ENERGY.— (1) CHIEF DIGITAL RECRUITING OFFICER.—Not later than 270 days after the date of the enactment of this Act, the Secretary of Energy shall designate a chief digital recruiting officer to oversee a digital recruiting office to carry out the responsibilities set forth in paragraph (2). (2) RESPONSIBILITIES.—The chief digital recruiting officer designated under paragraph (1) shall be responsible for— (A) identifying Department of Energy needs for specific types of digital talent; (B) recruiting technologists, in partnership with Department programs, by attending conferences and career fairs, and actively recruiting on college campuses; (C) integrating Federal scholarship for service programs into civilian recruiting; (D) offering recruitment and referral bonuses; and (E) partnering with human resource teams in Department programs to use direct-hire authorities to accelerate hiring. (3) STAFF AND RESOURCES.—The Secretary of Energy shall ensure that the chief digital recruiting officer designated under paragraph (1) is provided with staff and resources sufficient to maintain an office and to carry out the responsibilities set forth in paragraph (2).

Requires the Secretary of Defense, the Director of National Intelligence, and the Secretary of Energy each to designate a chief digital recruiting officer within 270 days of enactment. Each officer must identify digital talent needs, actively recruit technologists, integrate scholarship-for-service programs, and partner with human resource teams to accelerate hiring using direct-hire authorities. The section also requires each agency head to provide sufficient staff and resources for the digital recruiting office.

Compliance actions 3 items
1
The Secretary of Defense must designate a chief digital recruiting officer within 270 days to identify AI talent needs, recruit technologists, integrate scholarship-for-service programs, and accelerate hiring using direct-hire authorities.
2
The Director of National Intelligence must designate a chief digital recruiting officer within 270 days to identify intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5) AI talent needs, recruit technologists, offer bonuses, and accelerate hiring using direct-hire authorities.
3
The Secretary of Energy must designate a chief digital recruiting officer within 270 days to identify Department AI talent needs, recruit technologists, offer bonuses, and accelerate hiring using direct-hire authorities.
Sec. 5
DoD pilot program on AI development and prototyping fund
Government

(a)–(b) 4 ESTABLISHMENT OF PILOT PROGRAM AND ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) DEVELOPMENT AND PROTOTYPING FUND.— (1) PILOT PROGRAM.—The Secretary of Defense shall establish and carry out a pilot program to assess the feasibility and advisability of establishing a fund for the purpose set forth in subsection (b). (2) ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) DEVELOPMENT AND PROTOTYPING FUND.—Under the pilot program, the Secretary shall establish a fund to be known as the ''Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Development and Prototyping Fund'' (in this section referred to as the ''Fund''). (b) PURPOSE.—The purpose of the pilot program and the Fund is to support operational prototyping and speed the transition of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1)-enabled applications into both service-specific and joint mission capabilities with priority on joint mission capabilities for Combatant Commanders.

(c)–(g) 4 MANAGEMENT OF THE FUND.—The Fund shall be managed by the Secretary, acting through the Joint Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Center, in consultation with the Under Secretary of Defense for Research and Engineering, the Joint Staff, and the military services. (d) USE OF FUNDS FOR DEVELOPMENT OR PROTOTYPING.— (1) IN GENERAL.—The Secretary, acting through the Joint Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Center, may transfer such amounts in the Fund as the Secretary considers appropriate to a military service for the purpose of carrying out a development or prototyping program selected by the Secretary for the purpose described in subsection (b). (2) TREATMENT OF TRANSFERRED FUNDS.—Any amount transferred under paragraph (1) shall be credited to the account to which it is transferred. (3) SUPPLEMENT, NOT SUPPLANT.—The transfer authority provided in this subsection is in addition to any other transfer authority available to the Secretary or the Under Secretary. (e) CONGRESSIONAL NOTICE.— (1) NOTICE REQUIRED.—The Secretary shall notify the congressional defense committees of all transfers under subsection (d). (2) CONTENTS.—Each notice under paragraph (1) shall specify the amount transferred, the purpose of the transfer, and the total projected cost and estimated cost to complete the program to which the funds were transferred. (f) TERMINATION.—The pilot program and the Fund established under subsection (a) shall both terminate on the date that is two years after the date of the enactment of this Act. (g) AUTHORIZATION OF APPROPRIATIONS.—There are authorized to be appropriated $50,000,000 to be deposited into the Fund and to carry out this section.

Establishes a two-year pilot program under which the Secretary of Defense creates the Artificial Intelligence Development and Prototyping Fund, managed through the Joint Artificial Intelligence Center. The Fund supports operational prototyping and transition of AI-enabled applications into military capabilities, with priority on joint mission capabilities for Combatant Commanders. The Secretary may transfer amounts from the Fund to military services and must notify congressional defense committees of all transfers. Authorizes $50 million in appropriations.

Compliance actions 1 item
4
The Secretary of Defense must establish a two-year pilot program and $50 million AI Development and Prototyping Fund, managed through the Joint Artificial IntelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) Center, to accelerate transition of AI-enabled applications into military capabilities, and must notify congressional defense committees of all fund transfers.
Sec. 6
DoD resourcing plan for digital ecosystem
Government

(a)–(b) 5 IN GENERAL.—Within one year after the date of the enactment of this Act, the Secretary of Defense shall develop a plan for the development of a modern digital ecosystem that embraces state-of-the-art tools and modern processes to enable development, testing, fielding, and continuous update of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1)-powered applications at speed and scale from headquarters to the tactical edge. (b) CONTENTS OF PLAN.—At a minimum, the plan required by subsection (a) shall include— (1) an open architecture and an evolving reference design and guidance for needed technical investments in the proposed ecosystem that address issues including common interfaces, authentication, applications, platforms, software, hardware, and data infrastructure; (2) a governance structure, together with associated policies and guidance, to drive the implementation of the reference throughout the Department on a federated basis; and (3) a review of relevant authorities to operate and all information technology policies that may impede progress of a digital ecosystem.

Requires the Secretary of Defense, within one year of enactment, to develop a plan for a modern digital ecosystem supporting development, testing, fielding, and continuous update of AI-powered applications from headquarters to the tactical edge. The plan must include an open architecture and reference design, a governance structure with policies and guidance, and a review of authorities and IT policies that may impede progress.

Compliance actions 1 item
5
The Secretary of Defense must develop, within one year, a plan for a modern digital ecosystem to enable AI-powered application development, testing, and fielding at scale, including an open architecture, governance structure, and review of impeding IT policies.
Sec. 7
Accredited assessment of artificial intelligence systems
Government

(a)(1)–(2) 6 ACCREDITATION ASSESSMENT FOR ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) TESTING ORGANIZATIONS.— (1) IN GENERAL.—Not later than 540 days after the date of the enactment of this Act, the Director of the National Institute of Standards and Technology shall establish a program under which the Director provides accreditation to organizations that the Director determines competent at evaluating the potential, direct privacy, civil rights, and civil liberties effects of artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) systems used by covered agencies on United States personsUnited States personThe term "United States person" does not include a member of the Armed Forces or an employee of the Department of Defense.Sec. 3(6). (2) LIST.—The Director shall maintain a list of organizations accredited under the program established under paragraph (1).

(b) COVERED AGENCIES.—For the purposes of this section, a covered agency is— (1) the Department of Defense; (2) an element of the intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5) (as defined in section 3 of the National Security Act of 1947 (50 U.S.C. 3003)); or (3) the Federal Bureau of Investigation.

(c)(1)–(2) 7 ARTIFICIAL INTELLIGENCEArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) SYSTEM CRITERIA AND TESTING.— (1) IN GENERAL.—The privacy and civil liberties officers of covered agencies shall establish criteria for when an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system of a covered agency warrants accredited testing for privacy, civil liberties, and civil rights implications for individuals in the United States. (2) ADOPTION BY AGENCIES.—The head of a covered agency shall— (A) adopt the criteria established under paragraph (1); and (B) submit each artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system of the covered agency that meets the criteria for testing established under paragraph (1) to an organization accredited under the program established under subsection (a)(1) for evaluation in accordance with subsection (e).

(d)(1) 7 TESTING.— (1) IN GENERAL.—The head of a covered agency shall submit an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system that meets the criteria for testing established under subsection (c)(1) to an organization for an assessment of the performance of the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system according to the risk management framework developed under section 22A(c) of the National Institute of Standards and Technology Act (15 U.S.C. 278h–1(c))— (A) before procuring, fielding, or using the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system; and (B) after the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system passes an evaluation performed by an organization under subsection (e).

(e)(1)–(4) 7 SCOPE OF TESTING.—Each organization accredited under the program established under subsection (a)(1) shall, in evaluating an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system of a covered agency and as a condition on maintenance of such accreditation— (1) utilize metrics relevant to the mission and authorities of the covered agency; (2) develop approaches to test— (A) a software product used in an artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system, as installed in a test facility; and (B) cloud-based services relevant to the artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) system; (3) engage in binding agreements that enable the covered agency and other stakeholders to share confidential and proprietary data with the organization; and (4) collaborate with the covered agency to reach consensus on appropriate protocols and approaches for handling data, test results, and analyses.

(f) 8 ANNUAL REPORTS.—Once the Director of the National Institute of Standards and Technology establishes the program required by subsection (a)(1), each privacy and civil liberties officer of a covered agency shall, not less frequently than once each year, submit to Congress a report on the results of the testing on artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) systems of the covered agency under this section to ensure Congress understands the potential, direct privacy, civil rights, and civil liberties effects of such artificial intelligenceArtificial intelligenceThe term "artificial intelligence" includes the following: (A) Any artificial system that performs tasks under varying and unpredictable circumstances without significant human oversight, or that can learn from experience and improve performance when exposed to data sets. (B) An artificial system developed in computer software, physical hardware, or other context that solves tasks requiring human-like perception, cognition, planning, learning, communication, or physical action. (C) An artificial system designed to think or act like a human, including cognitive architectures and neural networks. (D) A set of techniques, including machine learning, that is designed to approximate a cognitive task. (E) An artificial system designed to act rationally, including an intelligent software agent or embodied robot that achieves goals using perception, planning, reasoning, learning, communicating, decisionmaking and acting.Sec. 3(1) systems.

This is the bill's core AI governance and accountability provision. It directs the Director of NIST to establish, within 540 days of enactment, an accreditation program for organizations competent at evaluating the privacy, civil rights, and civil liberties effects of AI systems used by covered agencies (DoD, intelligence community, FBI) on United States persons. Covered agency privacy and civil liberties officers must establish criteria for when an AI system warrants accredited testing, and agency heads must adopt those criteria and submit qualifying systems for accredited evaluation before procuring, fielding, or using them. Accredited organizations must utilize mission-relevant metrics, test both installed software and cloud-based services, enter binding data-sharing agreements, and collaborate on appropriate data-handling protocols. Each covered agency's privacy and civil liberties officer must submit annual reports to Congress on AI system testing results.

Compliance actions 3 items
6
The Director of NIST must establish, within 540 days, an accreditation program for organizations competent at evaluating the privacy, civil rights, and civil liberties effects of AI systems used by covered agencies on United States personsUnited States personThe term "United States person" does not include a member of the Armed Forces or an employee of the Department of Defense.Sec. 3(6), and must maintain a public list of accredited organizations.
7
Covered agencies (DoD, intelligence communityIntelligence communityThe term "intelligence community" has the meaning given such term in section 3 of the National Security Act of 1947 (50 U.S.C. 3003).Sec. 3(5), FBI) must establish criteria for when AI systems warrant accredited privacy, civil rights, and civil liberties testing, adopt those criteria, and submit each qualifying AI system to a NIST-accredited organization for evaluation before procuring, fielding, or using the system.
PS-01.2
8
Each covered agency's privacy and civil liberties officer must submit an annual report to Congress on the results of accredited AI system testing, covering the potential privacy, civil rights, and civil liberties effects of the agency's AI systems.
R-02.1

Passage Likelihood

Failed
Status Failed
Final action Read twice and referred to the Committee on Armed Services.

Legislative History

2021-05-19 Read twice and referred to the Committee on Armed Services.

Entry Last Reviewed

2026-05-16
AI generated