Hassabis Proposes Establishing an AI Regulatory Body

The Decoder · rss · 2026-07-14

DeepMind CEO Demis Hassabis believes that when it comes to advanced AI, "no one knows what will happen next." Therefore, the most realistic approach right now is "cautious optimism" and establishing guardrails first.

He proposed creating a US-led standard agency similar to the financial industry's FINRA. This body would comprise independent experts and open-source representatives to design evaluation protocols for frontier models and coordinate a slowdown in AI development if needed. Under this proposal, startups and research models could be exempted.

Original post →

More from AGI Musings

AGI Musings channel →