The meeting marks a thaw in a months-long standoff. Anthropic had refused the Pentagon unrestricted access to its Claude models over concerns about autonomous weapons and surveillance, prompting President Trump to order federal agencies to sever ties and label the firm a national security threat. Anthropic challenged the directive in court with mixed results; some agencies won permission to evaluate Mythos despite the ban. Treasury and State Department, which had terminated Anthropic products, now seek guidance on using Mythos for cyberdefense.
Mythos's ability to detect critical vulnerabilities has drawn urgent interest from tech and financial firms and international attention from the EU. The shift signals the administration is recalibrating its approach to balance AI innovation against national security concerns and deployment risks. Attorneys tracking federal AI policy should monitor the OMB safeguards framework and any formal agreements governing agency access to Mythos, as these will likely shape how other AI developers navigate government relationships going forward.