Comment by Omer Bilgin

In AI Policy Bulletin, we make the case for an international AI incident network. The same models are deployed in many countries at once, so a harm in one is often a preview of what is likely to come in others, but only if countries pass relevant warnings onto one another. We set out what such a network could do and a realistic way it could start: a bilateral agreement or small coalition of willing governments. Even two governments formally exchanging incident data on a shared template would be more coordination than what we have today.