The National Security Agency (NSA) has reportedly begun utilizing Anthropic’s Mythos AI model for offensive cyber operations, particularly targeting nations like China and Iran. This development suggests a shift in dynamics as Defense Secretary Pete Hegseth faces increasing challenges in his efforts against the AI company.
According to the Financial Times, approximately six engineers from Anthropic are working with the NSA, although their direct involvement in operations against adversaries remains uncertain. These engineers are primarily focused on tailoring AI models for specific purposes. This collaboration follows a contentious relationship, marked by Hegseth's campaign against Anthropic, which he labeled a supply chain risk back in March due to its refusal to adjust restrictions on its Claude model.
Despite Hegseth's attempts to blacklist Anthropic, there are indications that tensions may be easing. Recent visits by CEO Dario Amodei to the White House, including a meeting in mid-April, suggest a willingness from Anthropic to cooperate with military interests. This shift could enhance Anthropic's prospects as it prepares for an initial public offering, aligning its interests with those of the U.S. military and intelligence agencies.