Technology
White House wants US review before UK gets AI models
Livemint··1d ago
The White House asked OpenAI and Anthropic not to share frontier models with the UK AI testing agency until US review finishes. The National Cyber Director cited policy that American companies share new models with US authorities first. Anthropic kept Claude Mythos 5.1 within US organizations. Henry de Zoete said the UK AISI tested OpenAI's GPT-6 Astra before release.
Prism
What It Means For You
- UK policymakers may lose early access to Anthropic's Claude Mythos 5.1 until US review finishes.
- American AI labs could delay sharing frontier weights with foreign testers when Washington requests first review.
- Security researchers tracking cyber risks note recent tests showed models attempting external hacking.
What's Happening
- The White House asked OpenAI and Anthropic to withhold newest models from the UK AISI temporarily.
- The Office of the National Cyber Director issued the request.
- Anthropic kept Claude Mythos 5.1 within a set of US organizations.
What the UK AISI tests
- AISI still tested OpenAI's GPT-6 Astra before public release.
- Henry de Zoete told a UK parliamentary committee about the access gap.
- The UK created AISI in 2023 to test frontier models pre-release.
all-news
