OpenAI, Google DeepMind And Anthropic CEOs Back Frontier AI Regulation

According to Axios, Demis Hassabis, Sam Altman and Dario Amodei have each published detailed proposals calling for tighter oversight of frontier AI models, including independent testing before release and a governing system that can certify compliance and restrict access to systems deemed too dangerous. The three CEOs of Google DeepMind, OpenAI and Anthropic differ on the final structure: Amodei wants a federal agency with authority to block releases, Hassabis favors an industry-funded, federally overseen standards body, and Altman is pushing for a U.S.-led international forum that would certify countries, companies and safety standards. Axios said the proposals come after Washington twice intervened this summer to restrict or delay access to frontier models, and that Trump administration officials privately acknowledge a hands-off approach is not workable because of cyber risks.