6 mei 2026 · 13:46

US AI Pre-Deployment Testing: Google, Microsoft & xAI

Google, Microsoft and xAI have agreed to give the US government pre-deployment access to their frontier AI models for safety evaluations, marking a concrete shift from voluntary pledges to gated state oversight. The deal, announced through NIST's Center for AI Standards and Innovation, arrives just as European MEPs are pressing the Commission to secure equivalent access following the Mythos hacking revelations. This episode unpacks what the new US testing regime means for EU regulators, European AI developers and the critical infrastructure operators caught in the middle.