The Senate’s version of the National Defense Authorization Act includes a provision that bars artificial intelligence from launching nuclear weapons or using lethal force without direct human approval. The language, tucked into a must-pass defense policy bill, signals a growing push on Capitol Hill to rein in AI in the military before it outpaces existing rules.
What the ban covers
Specifically, the measure prohibits the Department of Defense from using AI to make the final decision on nuclear strikes or any lethal action. A human must remain in the loop — and in control — before a weapon is fired. The provision doesn’t ban AI from helping with targeting or intelligence analysis, but it draws a clear line at the launch button.
Why lawmakers moved now
Congress has been wrestling with how to govern AI in warfare as the technology gets faster and more autonomous. The NDAA provision is one of the first concrete legislative steps. It prioritizes ethical standards and human accountability, according to the bill’s backers. The move also reflects concerns that AI could misinterpret data or escalate conflicts unpredictably if left unchecked.
What happens next
The Senate bill must still be reconciled with the House version, which may have different AI language. The final NDAA is expected to pass later this year. Until then, the Pentagon is operating under existing guidelines — and watching closely to see how far Congress will go.
