Anthropic Refuses to Grant UK AI Safety Agency Access to Test Latest Models

TechFlow reports that on September 9, according to Sina Finance, Anthropic has refused to submit its latest model, Claude Mythos 5.1, to the UK Artificial Intelligence Safety Institute (AISI) for testing prior to release. The model is exclusively accessible to vetted U.S. institutions. This decision has raised alerts within Whitehall and internally at AISI, with UK national security officials concerned it will set a precedent that leaves the AISI unable to access cutting-edge models in the future.

Analysts note that this move may be linked to the U.S. government's increasingly stringent export control policies on the AI industry. Previously, the U.S. government imposed export controls on Anthropic's Mythos and Fable models, prohibiting foreign nationals from accessing them. Anthropic declined to comment, stating only that it is "coordinating with the U.S. government to expand model accessibility to more domestic and international partners as soon as possible."