DeepMind proposes support from international AI model review agencies Musk and Altman Tables

On July 17, Hassabis, CEO of Google DeepMind, proposed to establish an international independent agency to conduct unified testing and review of cutting-edge AI models before release, and plans to go to Washington next week to discuss relevant plans with the U.S. government. Hassabis said that the advanced cyber attack capabilities of Anthropic's Mythos model highlight the need to strengthen AI security supervision. The plan proposes that independent experts will be responsible for review, leading AI companies will provide funding, models will be tested for up to 30 days, and coordinate with the industry to slow down cutting-edge model research and development if necessary. OpenAI CEO Altman and Musk have publicly expressed support.