Artificial Intelligence Safety: Contributions from the Nuclear Community

Join us for a discussion with Christopher Hanson, White Flint Partners LLC, and former Chair, U.S. Nuclear Regulatory Commission on Artificial Intelligence Safety: Contributions from the Nuclear Community.

In 2024, Sam Altman, CEO of OpenAI, said "We have called for the creation of a federal agency that can grant licenses to create AI models above a certain threshold of capabilities... akin to how operators of nuclear facilities are required to be licensed." Roge Karma wrote in August in The Atlantic, “What humanity needs is an international AI nonproliferation treaty.”  All the while, nuclear safety, safeguards, and security practitioners have been quietly muttering to themselves, “Who knows how to guard against low-probability, high-consequence events? Who understands the high value of transparency in both engineered and human systems? Who has developed efficient and effective systems for safety and security oversight and verification? Us, that’s who!” This discussion will begin by identifying high level principles and practices from the nuclear “three S” world and how those might inform the current domestic and international debate on AI safety. Examples of current approaches to AI safety (the United Kingdom, California) will be highlighted to see where nuclear three S principles can inform or enhance subsequent iterations. The seminar is expected to be highly interactive with the lofty goal of converging on a handful topics where experience in nuclear translates well to the current AI safety debate.