Sponsors
The bill strengthens human control, safety testing, and limits on domestic surveillance by DoD AI—reducing automation risks (including for nuclear decisions)—but it also grants a two-year waiver power concentrated in the Secretary, with short congressional notice and compliance costs that could enable temporary deployments and complicate interagency operations.
U.S. service members and the public: the bill bars automation of nuclear launch decisions so AI cannot initiate or execute nuclear strikes without human control.
Service members: the bill prevents fully autonomous lethal weapon employment, keeping human judgment in life-or-death targeting decisions.
Americans: the bill restricts DoD domestic uses of AI for surveillance or profiling, reducing risks of targeting or monitoring without individualized legal basis.
Service members and the public: the Secretary of Defense can waive the ban on autonomous lethal use for up to two years, permitting temporary deployment of systems without continuing congressional approval.
Taxpayers and service members: the waiver authority is non-delegable and concentrates decision-making in a single official, raising risks if that authority is misused or continuity is broken.
Congress and the public: the five-day notification window may be too brief for meaningful congressional review of complex or classified justifications before systems are fielded or modified.
Based on analysis of 4 sections of legislative text.
Restricts DoD AI use by banning AI-controlled nuclear launch, limiting domestic surveillance without individualized legal basis, and prohibiting lethal autonomous weapons absent required human judgment, with a limited waiver and rapid congressional notice.
Limits how the Department of Defense can use artificial intelligence by banning AI-controlled nuclear launch, restricting domestic monitoring or targeting without an individualized legal basis, and prohibiting lethal autonomous weapon use without required human judgment. It expresses a nonbinding preference for aggressive, secure AI adoption consistent with a recent national AI plan, defines AI by federal statute, and creates a narrow, time-limited waiver process the Secretary of Defense (not delegable) can use with rapid congressional notification and specified disclosures.
Introduced March 17, 2026 by Elissa Slotkin · Last progress March 17, 2026