An AI model called Mythos, built by Anthropic, was let loose on the US government's most sensitive computer systems under a restricted program called Project Glasswing. According to Senator Mark Warner, who disclosed the results at a June 11 hearing, NSA chief General Joshua Rudd told him the model "broke into almost all of our classified systems, not in weeks, but in hours." An AP report caveats that identifying vulnerabilities isn't the same as fully exploiting them — but the speed alone has rattled officials. (Euronews, Reuters)

The revelation landed days after the Five Eyes intelligence alliance — the US, UK, Canada, Australia, and New Zealand — issued a rare joint statement warning that frontier AI models will "fundamentally transform both offensive and defensive cyber capabilities," and that the timeline for devastating attacks is "not years, it is months." The statement urged leaders to "act swiftly," while also positioning AI as the best defense against AI-powered threats. (ABC News, Reuters)

What makes this a zeitgeist moment isn't just the capability — it's the fight it triggered. Anthropic refused to let the US military use its models for domestic surveillance or autonomous weapons. The Trump administration retaliated by blacklisting the company and suspending worldwide access to Mythos and its sibling model Fable. Over 100 cybersecurity experts from companies including Adobe and Nvidia signed a letter urging the government to lift the restriction, arguing Mythos is "not uniquely good" at offensive tasks and that gutting the best defensive tool helps nobody but adversaries. The AI that found America's vulnerabilities is now the AI America can barely touch.