The US let Britain test American AI models before release again.
The Telegraph reported on 11 October that the White House reversed a request it made last month. It had asked developers such as OpenAI and Anthropic not to show Britain's AI Security Institute new models before launch. Neither the White House nor the two companies commented to Reuters.
Why it mattersThe institute is one of the few outside bodies that tests frontier models before the public gets them. Its reports are often the first independent look at what a new model does.
Read in the app — free, in 9 languages
Related stories
Britain's AI minister will say firms cannot police their own models.2026-10-12Activision took down Call of Duty browser ports built with AI agents.
2026-10-12Japan's ruling party says the country is in a cyber emergency.
2026-10-11Tech firms shape Medicare's AI app rules in a 1,700-member chat.
2026-10-11The fixer in a billion-dollar Nvidia smuggling case pleaded guilty.
2026-10-10