What happened: Demis Hassabis, CEO of Google DeepMind, published an op-ed calling for an independent international AGI oversight body — funded by labs, governed by technical experts and civil society, with audit authority over frontier model deployments.
Why it matters: This is the first time a major lab CEO has explicitly endorsed external oversight with teeth. The previous position was "trust our internal safety teams." The shift: labs can't grade their own homework when the exam is existential risk.
The proposal
- International AGI Observatory: Multilateral body with technical audit powers
- Mandatory pre-deployment review: Frontier models (>10^25 FLOPs) assessed before release
- Incident reporting: Standardized disclosure for capability jumps, alignment failures, misuse
- Funding: 0.1% of lab compute budgets pooled into the observatory
What we're watching
Whether other lab CEOs (Altman, Amodei, Sutskever) endorse or counter-propose. The EU AI Act and US AI Safety Institute are both watching this conversation — the first concrete proposal may become the regulatory template.
Originally published on ayraix.com, practical AI for enterprise builders.
Top comments (0)