Google DeepMind CEO Hassabis urges rules for AI before AGI
Demis Hassabis, the CEO of Google DeepMind, is urging governments and tech leaders to act fast and set up rules for advanced AI before we reach artificial general intelligence (AGI): basically, AI that thinks like humans.
He says there's a "precious window" right now to make sure things stay safe as AI gets smarter.
Hassabis proposes U.S.-led AI model reviews
Hassabis suggests forming a U.S.-led coalition to review new AI models before they're released.
He imagines an oversight group (like FINRA in finance) with experts and policymakers testing for risks like cyberattacks or bioweapons.
The idea is catching problems early so people stay protected.
Hassabis hopes to launch initiative 2026
His proposal is getting positive feedback from US officials, top AI labs, and European leaders.
With rapid changes in the field, many agree it's time for stronger oversight, and Hassabis hopes to launch this initiative by the end of 2026.