Google DeepMind chief Demis Hassabis called for Washington to establish a federally overseen, U.S.-led standards body to test “frontier” AI models for national-security risks, including cyber and bio threats. Framing the effort as a public-private partnership akin to FINRA, he said such a watchdog should be well funded, staffed with independent experts, and initially receive models on a voluntary basis before moving to mandatory pre-release reviews for the U.S. market. The push follows mounting tensions over government curbs on advanced AI deployments and comes as Chinese developers roll out competitive systems that are gaining traction with U.S. firms. Hassabis argued America’s economic and technical heft positions it to set global benchmarks as lawmakers weigh how to manage the U.S.-China AI race.
Related articles:
AI Risk Management Framework (AI RMF 1.0) | NIST
OECD AI Principles
European approach to Artificial Intelligence | European Commission
