UK secures AI model access for safety testing

UK AI Minister Kanishka Narayan announced that Britain's AI Security Institute (AISI) is conducting safety evaluations of the latest AI models, including Gemini 4 Argon, before public release. This follows reports that the US had asked OpenAI and Anthropic to withhold models from UK testing. Narayan also revealed AI tools identified over 1,000 government system vulnerabilities.