Hassabis reportedly proposed an independent frontier-AI oversight body

AI safetyUnited States / United Kingdom13 August 2026

Organisation: Google DeepMind. Status: reported proposal, not an adopted government programme.

The it was reported that Demis Hassabis proposed an industry-funded oversight organisation involving government agencies, national laboratories and AI developers to evaluate frontier systems for security risk.

Why it matters and impact: A credible shared testing body could reduce dependence on self-attestation and support common evaluation baselines. Governance, funding, legal authority and international participation remain unresolved.