FT says Anthropic withheld a model from the UK AI Security Institute, as a researcher puts extinction risk above 10 percent within the next decade

Sources: BBC report, discussion
Summary: The BBC carries a Financial Times account that Anthropic withheld its latest model from the UK AI Security Institute, and states Anthropic declined to comment on it. No source in the report names that model, and a Cabinet Office spokesperson declined to say whether a model had been withheld, stating the government continues to collaborate closely with industry partners, including Anthropic, to make models safer. The same report states that Evan Hubinger warned there is a greater than 10 percent chance AI could kill all humans within the next decade, in reply to Jacob Coxon, who has just left Anthropic and previously worked at OpenAI.
Why it matters: The withheld-model account is the testable part, because pre-deployment access for the national security institute is either granted or it is not, and neither party has said which.
Follow-up: Track whether Anthropic or the UK AI Security Institute confirms the withheld-model account.