Sponsors
The bill strengthens safeguards, transparency, and civil‑liberty protections around AI‑enabled weapons but preserves a temporary waiver that could permit risky autonomous force use and may slow deployments and increase costs while complicating some domestic intelligence activities.
All Americans and military personnel: bans on AI control of nuclear launch and limits on domestic monitoring reduce the risk of catastrophic misuse and protect civil liberties.
Military personnel and commanders: clearer rules and required safeguards (error‑rate testing and mandated human supervision) for AI-enabled weapon systems reduce the likelihood of accidental engagements and improve operational safety.
Congress and the public: required timely, detailed notifications about any waived AI weapons (including rationale, tests, and safeguards) improve legislative oversight and transparency for high‑risk systems.
Military personnel and the public: the Secretary's authority to waive prohibitions on autonomous lethal force for up to two years allows temporary deployment of high‑risk systems that could cause harm or be misused.
Taxpayers and military programs: rigorous testing and reporting requirements could delay urgent deployments in crises and increase program costs.
Law enforcement and local government agencies: restrictions on domestic monitoring and profiling may limit some defense or interagency intelligence activities and complicate cooperation with domestic law enforcement.
Based on analysis of 4 sections of legislative text.
Restricts certain DoD uses of AI—especially autonomous lethal systems—and allows only narrow, time-limited waivers from the Secretary of Defense with fast congressional notice and comparative testing requirements.
Introduced March 17, 2026 by Elissa Slotkin · Last progress March 17, 2026
Prohibits certain Department of Defense uses of artificial intelligence and sets strict limits on using AI to employ lethal force by autonomous weapon systems. It defines AI by reference to existing federal law, endorses alignment with the July 2025 America’s AI Action Plan, and creates a narrow, time-limited waiver process that only the Secretary of Defense may grant with fast congressional notice and specific testing and documentation requirements. The waiver may last up to one year and may be renewed once after the Secretary certifies to congressional defense committees that extraordinary national security circumstances require it and that the system’s error rate is no worse than trained human operators under equivalent conditions. The Secretary must notify Congress within five days of issuing waivers for development, fielding, or major modifications and must include rationale, system details, safeguards, test comparisons to humans, and expected duration (unclassified with a classified annex if needed).