FT: NSA stations ~6 Anthropic engineers to ready Claude Mythos for cyber operations despite Pentagon 'supply-chain risk' ban

Citing Financial Times reporting, TechCrunch reported on June 5, 2026 that the US National Security Agency has stationed approximately six Anthropic engineers on-site to help the agency deploy Claude Mythos, Anthropic's frontier cybersecurity model, for use in cyber operations. The NSA, which operates under Department of Defense jurisdiction and conducts both intelligence collection and offensive cyberattacks, is integrating a model Anthropic has deliberately kept under restricted access precisely because of concerns it could be exploited to identify software vulnerabilities and chain them into attacks. The deployment is notable because it occurs despite a standing federal ban on using Anthropic technology: the Pentagon designated Anthropic a 'supply-chain risk' after the company refused to permit government use of its models for mass domestic surveillance and autonomous weapons.

The report does not confirm whether the engineers or Mythos are yet active in live NSA hacking operations. Both the NSA and Anthropic declined to comment. The story sits at the intersection of frontier-model capability and national-security policy, illustrating how the same vulnerability-discovery capability that powers Anthropic's defensive Project Glasswing initiative is simultaneously being courted for offensive state use — and how a vendor's stated safety red lines (no surveillance, no autonomous weapons) collide with intelligence-agency demand.

It also underscores the geopolitical sensitivity around Mythos access, which Anthropic has been rationing globally.