OpenAI, Anthropic, Google in talks to create AI safety standards body

OpenAI, Anthropic, and Google are in discussions to establish an industry standards body for testing and auditing advanced artificial intelligence models, a move aimed at addressing growing safety concerns. The initiative, which has been underway since at least July, originated from a proposal by Google DeepMind founder Demis Hassabis. OpenAI confirmed the talks on Tuesday, with its global policy chief Chris Lehane stating the company is working with its rivals on steps to address AI safety. He added that OpenAI does not believe an antitrust waiver is needed for the three companies to coordinate on safety matters. The proposal envisions a U.S.-centered body modeled on the Financial Industry Regulatory Authority (FINRA), which oversees Wall Street. Hassabis outlined the idea in a July essay, suggesting a public-private partnership funded by industry but staffed with independent technical experts to evaluate the most powerful AI systems before their public release. The discussions have continued even as the broader industry remains divided.