In Geneva on July 6, the UN convened its first Global Dialogue on AI. It comes with an independent scientific report by 40 experts, led by Yoshua Bengio and Maria Ressa, and the finding is clear: over a billion people use conversational AI every week, and reliable methods for controlling autonomous systems are still missing.
- The UN is holding the first Global Dialogue on AI governance (Geneva, July 6-7) with an independent scientific report from 40 experts.
- Findings: 1 billion+ weekly users; US 75% / China 15% of top AI compute; no guarantee that systems won't violate their instructions.
- Read: control gets built in (gates, least privilege), not waited on from outside - the system, not trust in the model.
For the first time, the UN sits down specifically to tackle the question of who governs artificial intelligence, with a scientific report on the table. And the tone is more worried than ceremonial.
One sentence carries the whole report, and if you're looking for a guarantee in it, you won't find one: 'no guarantee that the system won't violate the instructions given to it'. The rest is protocol. Photos for the press. The big risk from AI isn't news - we've known it for a while. What matters here is different: the highest podium in the world put it on the record. Guterres said it without dancing around it - 'we can no longer say we didn't know'. With that, 'we didn't know' stops working as an excuse for anyone.
Bengio puts his finger on the wound: capabilities are outrunning both understanding and the speed at which governments react. The real gap isn't between knowing and not knowing. It's between the machine's tempo and the institution's tempo. The document acknowledges the gap. It hasn't closed it yet. From here on, every delay is a choice, not a surprise. And when AI overrides someone's instruction, nobody at the table in Geneva gets to shrug.