Google WANTS an AI "IAEA." Yes, really.
What happened
Google DeepMind CEO Demis Hassabis has proposed the creation of a global regulator for frontier AI, modeled loosely on international institutions that oversee technologies with civilization-level consequences.
Why this matters
This is not a random policy white paper from a think tank. It is the CEO of one of the labs most likely to build the next generation of frontier AI saying voluntary promises may no longer be enough.
The slightly cynical read
Calling for regulation can also be a strategic move. The companies already at the frontier can afford expensive audits, safety evaluations and compliance teams. Smaller challengers may not.
What to watch next
Whether the proposal gets support from Washington, whether OpenAI and Anthropic publicly align with it, and whether "global" really means global — or mostly an American club with international branding.