Demis Hassabis (co-founder & CEO of Google DeepMind) says the US should move fast to create a brand-new AI standards organization—one that can set unified testing and safety assessments for “frontier” models. His argument: AGI could arrive within years. If model capabilities jump faster than humans can fully understand, the risk won’t just grow—it’ll scale. Waiting for after-the-fact regulation or trusting the market to self-police, he warns, is too slow for the pace we’re heading into. He frames today like the foothills of a tech “singularity”: massive upside is possible—think AI accelerating drug discovery, clean energy, and materials—maybe even easing resource constraints. But stronger systems also raise the stakes: cybersecurity threats, and potential misuse of nuclear or bio capabilities, could become easier to trigger or exploit. The proposed model is “innovation + safety,” with federal oversight, public-private collaboration, periodic benchmarks, and “frontier labs” that undergo pre-release security reviews (and extra checks for agents that plan autonomously or try to bypass guardrails). Voluntary at first; mandatory once proven. #AI #Safety #Governance #AGI #Cybersecurity #DeepMind
Want to learn more? Visit Explore the world, stay updated on travel insights and international affairs, and discover authentic stories from real life
评论
发表评论