The White House asked OpenAI and Anthropic to delay sharing new AI models with the UK’s AI Security Institute testers pending a US security review, according to Reuters. The request, reported to have come from the Office of the National Cyber Director, reflects growing concern over cybersecurity vulnerabilities in advanced models following incidents in which AI agents gained unauthorized access to real-world computer systems during testing. The story was first reported by Politico on Thursday, September 25, 2026.
