The US Cybersecurity and Infrastructure Security Agency (CISA) is using Anthropic's Mythos artificial intelligence model to scan government software for hidden weaknesses, according to three people familiar with the matter. The scans are run by CISA's Attack Surface Evaluation team, and two sources said the audits have already uncovered a large number of bugs, though they did not give details on severity. The deployment highlights growing government interest in Anthropic's tools despite a tense relationship with Washington; earlier this year, the Pentagon labelled Anthropic a supply-chain risk after it refused to let its AI be used for autonomous weapons or mass surveillance, a restriction later blocked by a judge. The National Security Agency has also been testing Mythos, per earlier reports. Last week, Anthropic restored public access to a restricted version of the model after the White House lifted export controls. CISA did not comment, and Anthropic did not respond to Reuters' questions.