Responsible Artificial Intelligence Defense Act of 2026
- Bill Number
- S. 4707
- Origin Chamber
- Senate
- Congress
- 119th Congress, Session 2
- Policy Area
- Armed Forces and National Security
- Status
- Introduced
- Latest Action
- 2026-06-08: Read twice and referred to the Committee on Armed Services.
- Last Updated
- 2026-07-01T20:04:50Z
AI-Generated Summary
Summary of S. 4707: Responsible Artificial Intelligence Defense Act of 2026
Purpose
This legislation amends title 10 of the United States Code to establish Department of Defense (DoD) policy on maximizing the use of autonomous systems and artificial intelligence (AI) capabilities in weapons and other applications. It requires rigorous human oversight to ensure compliance with the law of war, treaties, safety rules, rules of engagement, and protections for privacy and civil liberties. The bill also sets up review, verification, testing, and oversight processes for these technologies.
Key Provisions
- Policy Framework: Directs the DoD to maximize autonomy and AI use where practical, while mandating continuous human oversight and responsibility for development, deployment, and use.
- Human Judgment Levels: Requires categorization of systems into two risk-based levels (Level 1 for low risk; Level 2 for moderate to high risk to life, safety, or mission).
- Review and Verification Process:
- Pre-prototyping or development review by the Under Secretary of Defense for Research and Engineering, including design checks for human judgment, target engagement limits, risk mitigation, cybersecurity, and legal compliance.
- Pre-fielding review (with Joint Chiefs input) confirming operational readiness, safety features, updated legal reviews, and monitoring plans.
- Verifications are valid for three years, with provisions for substantially similar variants.
- Privacy and Legal Safeguards: Mandates privacy impact assessments for certain AI capabilities; requires coordination with the DoD General Counsel for legal reviews aligned with existing directives.
- Exceptions: Lists 10 categories exempt from policy and requirements, such as operator-supervised systems for materiel targets, non-lethal cyberspace tools, unguided munitions, and mines. Includes a notification process for certain high-autonomy systems deemed safer than alternatives.
- Prohibitions: Bans use for nuclear weapon launch decisions, unauthorized domestic surveillance or targeting without a warrant, and lethal force by autonomous systems lacking appropriate human judgment.
- Testing and Evaluation: Directs rigorous verification, validation, and testing by the Director for Operational Test and Evaluation, including cyber resilience, adaptability to new data, operator training, and ongoing monitoring for changes.
- Working Group: Establishes the Autonomy and Artificial Intelligence Working Group to advise on reviews and develop safety standards.
- Reporting: Requires annual reports to congressional defense committees through January 2037.
- Definitions: Clarifies "artificial intelligence," "autonomous weapon systems" (including operator-supervised variants), and "autonomy."
Significant Changes to Existing Law
- Introduces a new section (4577) in Chapter 345 of title 10, U.S. Code, creating mandatory planning, oversight, and verification processes that build upon and formalize aspects of prior DoD directives (e.g., 3000.09 on Autonomy in Weapon Systems).
- Establishes specific two-level human judgment frameworks and privacy assessments not previously codified in statute.
- Adds prohibitions on certain AI uses and a dedicated working group, shifting from directive-based guidance to statutory requirements.
- Extends oversight to AI capabilities beyond traditional weapons systems, with updated legal reviews required at key stages.
Potential Impacts
- Government Agencies: Increases DoD workload for reviews, testing, and legal/privacy assessments, involving the Under Secretary for Research and Engineering, Director for Operational Test and Evaluation, and privacy officials. May require additional resources for the new working group and annual reporting.
- Citizens: Strengthens protections against unauthorized domestic use of AI for surveillance, supporting privacy and civil liberties under existing laws like the Privacy Act.
- International Relations: Promotes adherence to the law of war and treaties, potentially influencing U.S. credibility in arms control discussions or alliances, while exceptions and waivers allow flexibility in conflicts.
Main Stakeholders Affected
- Department of Defense leadership, military departments, and operational commanders/operators.
- Congressional defense committees (oversight and notification recipients).
- DoD acquisition and testing personnel.
- U.S. persons whose privacy data may be involved in AI systems.
Notable Legal, Constitutional, or Political Implications
- Reinforces constitutional limits on domestic surveillance by requiring warrants based on probable cause for certain AI uses.
- Emphasizes legal reviews tied to established DoD directives on acquisition and the law of war, potentially reducing risks of unintended engagements or violations.
- Politically, it signals a push for responsible AI integration in defense, balancing technological advancement with accountability, and could affect debates on autonomous weapons internationally.
This summary was generated by AI and may contain inaccuracies. Refer to the official source document for the authoritative text.
Sponsor
Sen. Coons, Christopher A. [D-DE]
Cosponsors (1)
Recent Actions
- 2026-06-08: Read twice and referred to the Committee on Armed Services.
- 2026-06-08: Introduced in Senate
Bill Versions
- Responsible Artificial Intelligence Defense Act of 2026 — issued 2026-06-08 — PDF (22 pages)