Sponsors
The bill strengthens civil liberties and oversight by restricting DoD domestic AI surveillance and mandating human supervision and reporting, but it still permits time-limited waivers and contains validation and transparency gaps that could allow risky autonomous systems to be deployed.
All Americans (civilian population) are better protected from DoD AI-enabled bulk surveillance and profiling because the bill restricts DoD use of AI for domestic monitoring without individualized legal bases, reducing risks to First Amendment and privacy rights.
Service members and civilians face lower risk of AI-driven accidental lethal action because the bill requires appropriate human judgment and supervision for autonomous weapons and critical functions.
Taxpayers and their representatives gain stronger federal oversight because DoD must provide written certifications and 5-day notifications for waivers and major changes, enabling Congress to monitor risky autonomous systems more quickly.
Military personnel and taxpayers face increased risk because the Secretary can waive restrictions and field autonomous lethal systems for up to a year (renewable), potentially allowing deployment before long-term safety and ethical concerns are resolved.
Service members may still be exposed to significant harm because requiring AI error rates to not exceed documented human error rates is hard to validate and could permit risky systems if human baselines are high or poorly defined.
Taxpayers and the public may have limited real transparency because 5-day unclassified notifications paired with classified annexes can withhold key details, reducing public accountability for risky systems.
Based on analysis of 4 sections of legislative text.
Introduced March 17, 2026 by Elissa Slotkin · Last progress March 17, 2026
Limits Department of Defense use of artificial intelligence by banning AI control of nuclear launch, restricting AI for domestic monitoring/targeting without individualized legal basis, and prohibiting autonomous lethal force unless strict human oversight and a written, nondelegable Secretary of Defense waiver is granted. The bill endorses aggressive AI adoption in principle but requires detailed certification, rapid congressional notification, and performance evidence showing AI error rates do not exceed documented human error rates before certain waivers may be used.