US Department of Defense policy already prohibits AI from independently launching nuclear weapons. But amid growing fears of artificial intelligence driven by a plethora of potential threats, a bipartisan group of MPs decided to make sure it couldn’t.
“In all cases, the United States will maintain a person ‘in the loop’ for all critical actions to communicate and implement the President’s decisions to initiate and terminate the use of nuclear weapons.”
By the same token, the bill states that no independent regime without meaningful human oversight can launch a nuclear weapon or “select or engage targets” with the intent to launch one. The text states that “any decision to launch a nuclear weapon should not be taken by means of artificial intelligence.”
If this is already forbidden, why submit the bill? Its sponsors point out that the 2021 National Security Committee on Artificial Intelligence report recommended affirming the ban on autonomous nuclear weapons launches, not only to prevent this from happening within the US government but to spur similar commitments from China and Russia. The bill’s announcement brings attention to the potential risks of the current generation of autonomous AI systems, which is an ongoing concern in Congress and the tech world alike. And, as the press release noted, it provides an opportunity to highlight other sponsors’ nonproliferation efforts — such as a recent bill that restricts the president’s power to unilaterally declare nuclear war. I’ll let you make it clear War games Kidding yourself.