The NSA is using Anthropic's most powerful AI model Mythos

The NSA deployment isn't just a customer win — it's confirmation that Anthropic's calculated bet on a cybersecurity-specific model paid off faster than most observers expected, converting a political liability into a cleared-agency contract within days of Mythos Preview's release.

This story is the direct payoff of the dynamic we flagged on April 17, when The Verge reported that Anthropic's new cybersecurity model 'could get it back in the government's good graces' after weeks of friction with the Trump administration. That piece framed Mythos Preview as a rapprochement play; the NSA deployment confirms the play worked. It also sits in sharp tension with our April 16 coverage from MIT Technology Review, which noted Anthropic was in a legal dispute with the Pentagon over AI deployment in warfare — suggesting the company is simultaneously fighting one part of the national security establishment while winning contracts from another. The London expansion story from Wired, published the same week, now reads differently: geographic diversification may have been a hedge that Anthropic no longer needs as urgently.

Watch whether the Pentagon legal dispute over AI in warfare (covered by MIT Technology Review on April 16) gets quietly settled or dropped in the next 60 days — a resolution would signal that the NSA contract is part of a broader détente, not an isolated deal.