U.S. Sets Historic AI Policy for Nuclear Weapons in FY2025 NDAA, Ensuring Human Control
WASHINGTON, D.C. – Amid growing concerns about the role of automated systems in nuclear weapons, the U.S. has established its first policy governing the use of artificial intelligence (AI) in nuclear command, control and communications. Signed into law as part of the FY2025 NDAA, this historic measure ensures that AI will strengthen, rather than compromise, human decision-making in our nuclear command structure.
The policy allows AI to be integrated in early warning capabilities and strategic communications while maintaining human judgment over critical decisions like the employment of nuclear weapons, ensuring that final authorization for such consequential actions remains firmly under human control.
Through extensive engagement with Congress, Encode helped develop key aspects of the provision, Section 1638. Working with Senate and House Armed Services Committee offices, Encode led a coalition of experts including former defense officials, AI safety researchers, arms control experts, former National Security Council staff, and prominent civil society organizations to successfully advocate for this vital provision.
“Until today, there were zero laws governing AI use in nuclear weapons systems,” said Sunny Gandhi, Vice President of Political Affairs at Encode. “This policy marks a turning point in how the U.S. integrates AI into our nation’s most strategic asset.”
The bipartisan-passed measure emerged through close collaboration with congressional champions including Senator Ed Markey, Congressman Ted Lieu, and Congresswoman Sara Jacobs, establishing America’s first legislative action on AI’s role in nuclear weapons systems.
About Encode: Encode is a leading voice in responsible AI development and national security, advancing policies that promote American technological leadership while ensuring appropriate safeguards. The organization played a key role in developing California’s SB 1047, landmark state legislation aimed at reducing catastrophic risks from advanced AI systems. It works extensively with defense and intelligence community stakeholders to strengthen U.S. capabilities while mitigating risks.
