Anthropic

Anthropic’s Mythos model reportedly alarmed Washington after it cracked NSA classified systems during an authorized red-team exercise in hours. Senator Mark Warner said NSA and U.S. Cyber Command chief General Joshua Rudd told him Mythos had broken into almost all classified systems “in hours,” not weeks. The company had already admitted in April that Mythos was too effective at finding security holes for a public release, and it moved the model into a controlled Project Glasswing program with about 200 vetted partners. fudzilla.com Anthropic ban smells of panic – Fudzilla.com AI News Jun 23, 2026 by Nick Farrell Anthropic ban smells of panic Anthropic’s Mythos model reportedly scared Washington senseless after cracking NSA classified systems… fudzilla.com longtail-tier 2026-06-23 View full article →

That security scare lines up with the Trump administration’s 12 June order restricting Anthropic’s models to U.S. citizens only. Anthropic could not verify nationality at scale, so it shut the models off globally with 90 minutes’ notice, cutting off allies, researchers, and foreign-national staff. The ban was the first time the U.S. government had imposed export controls directly on an AI model rather than on the chips underneath it. fudzilla.com Anthropic ban smells of panic – Fudzilla.com AI News Jun 23, 2026 by Nick Farrell Anthropic ban smells of panic Anthropic’s Mythos model reportedly scared Washington senseless after cracking NSA classified systems… fudzilla.com longtail-tier 2026-06-23 View full article →