The Trump administration is finalizing a classified contract that will allow US intelligence agencies, led by the NSA, to access Anthropic’s advanced AI models-including the newly developed Mythos system. The deal includes a key restriction: the technology cannot be used on data involving US citizens.
This marks a reversal from February 2026, when President Trump ordered an immediate halt to federal use of Anthropic’s technology after the company refused to grant the Pentagon unrestricted access. At that time, the Pentagon designated Anthropic a supply chain risk, despite billions already invested across government networks.
Ongoing chip shortages had worsened operational gaps for spy agencies, prompting White House Chief of Staff Susie Wiles to authorize interim use of Anthropic’s models while negotiations continued.
By April 2026, Trump acknowledged a potential deal for Department of Defense use, noting Anthropic was “shaping up” to meet government expectations. The White House has committed $9 billion to boost AI infrastructure for defense. The prohibition on using Anthropic’s technology with US citizen data appears to be the compromise that sealed the agreement.