NSA reportedly using Anthropic's Mythos AI model despite Pentagon feud

At a glance:

• The National Security Agency is reportedly using Anthropic's Mythos Preview AI model despite the Trump administration's February order to stop using Anthropic services
• Anthropic announced Mythos Preview in April as a model "strikingly capable at computer security tasks"
• The company is currently in a legal battle with the US government after being labeled a "supply chain risk"

What happened

According to Axios, speaking with two sources with knowledge of the matter, the National Security Agency has been using Anthropic's new Mythos Preview AI model. This development comes despite the months-long feud between Anthropic and the Pentagon, which was triggered when the Trump administration ordered all government agencies to stop using Anthropic's services in February. The company had refused to budge on certain safeguards for military uses during contract talks, leading to the directive. The news about NSA's use of Mythos comes just days after Anthropic CEO Dario Amodei met with White House chief of staff Susie Wiles and other officials to reportedly discuss the model. The White House described the meeting as "productive and constructive," though President Trump claimed he had "no idea" about it when questioned by reporters.

Mythos Preview capabilities

Anthropic announced Mythos Preview at the beginning of April, describing it as a general-purpose language model with particular strengths in computer security. The company specifically highlighted that the model is "strikingly capable at computer security tasks," which appears to have made it attractive to the NSA and other organizations. According to Axios' sources, the NSA is one of approximately 40 organizations that Anthropic has given access to Mythos Preview. One source indicated that the model is "being used more widely within the department" as well, suggesting that its adoption may be expanding beyond initial testing phases. The development of specialized AI models for security purposes represents a growing trend in both private and government sectors, as organizations seek advanced tools to address increasingly sophisticated cyber threats.

The ongoing legal battle

Despite the apparent progress in relations with some parts of the government, Anthropic remains embroiled in a legal battle with the US Department of Defense. In March, the company filed lawsuits against the Pentagon in two separate courts after the Trump administration labeled it a "supply chain risk." This designation carries significant implications for government contracting and could potentially limit Anthropic's ability to work with federal agencies. The legal situation is complex: while Anthropic was granted a preliminary injunction by one court to temporarily block this designation, federal judges in the other court denied its motion to lift the label. This mixed legal outcome leaves the company in an uncertain position regarding its future relationship with the Department of Defense and other federal agencies.

Why this matters

The reported use of Mythos by the NSA represents a significant development in the relationship between Anthropic and the US government, particularly given the recent tensions. It suggests that despite the administration's public stance, there may be recognition of the value that Anthropic's technology can provide, especially in specialized domains like cybersecurity. This situation highlights the complex interplay between national security interests, technological innovation, and political considerations. For the AI industry, this case may set precedents for how government agencies engage with AI companies, particularly those developing technologies with dual-use potential. The ongoing legal battle will likely have implications for how AI companies navigate government contracts and national security considerations in the future.