DeepMind 希望建立 AI 版 FDA — 實驗室會自願加入嗎？

google deepmind ceo demis hassabis called for a new industry standards body to regulate frontier ai, modeled after the financial industry regulatory authority. the proposal suggests frontier labs voluntarily submit models for review 30 days prior to release, with the possibility of formalized mandatory deployment requirements in the us later. the regulator would be funded by the ai industry but backed by the government, aiming to fix the technical expertise gaps seen in recent federal reviews of anthropic's mythos and openai's sol. this approach attempts to satisfy white house concerns against an executive branch regulator while maintaining technical rigor. if implemented, it would shift ai safety enforcement from the government to a self-regulatory organization staffed by industry experts.