In a recent development, Senator Edward Markey, a Democrat from Massachusetts, is calling for the United States to pass legislation that would prevent artificial intelligence (AI) from being involved in nuclear power decisions.
AI and Nuclear Power: A New Kind of Threat
Senator Markey expressed his concerns about the militarization of increasingly powerful AI systems and the potential nuclear threat they pose. He emphasized the need for legislation to keep AI away from nuclear decisions, asserting that action must be taken before it’s too late.
Proposed Amendments to the National Defense Authorization Act
Markey’s office revealed that he has filed over a dozen amendments to the National Defense Authorization Act. These include language that would prohibit the use of AI in the U.S. military’s nuclear launch decisions. The amendments also aim to advance nuclear disarmament and nonproliferation, and propose a shift in federal spending from nuclear weapons development to vaccine research.
The Block Nuclear Launch by Autonomous Artificial Intelligence Act
One of the amendments is based on the Block Nuclear Launch by Autonomous Artificial Intelligence Act. This act would prohibit AI from making nuclear launch decisions. The legislation was introduced in April by Markey and bipartisan Representatives Ted Lieu, Don Beyer, and Ken Buck.
Keeping Humans in the Loop
Markey and his colleagues argue for the necessity of keeping humans in the loop when it comes to life or death decisions, especially concerning the use of deadly force like nuclear weapons. Their bill would codify existing Pentagon policy that requires a human to be involved in any decisions regarding the use of nuclear weapons.
Pentagon’s Budget for AI Research and Development
The Pentagon’s fiscal year 2024 budget calls for $1.8 billion solely for the research and development of AI capabilities. This highlights the increasing importance and potential of AI in military and defense strategies.
As AI continues to evolve and become more powerful, the discussion around its ethical use, especially in areas of high stakes like nuclear power, becomes increasingly critical. This proposed legislation is a step towards ensuring that AI is used responsibly and does not pose a threat to global security.