The relationship between the U.S. government and Anthropic has reached a state of high-stakes paradox. While the Trump administration has officially labeled the AI startup a "supply chain risk" and sought to bar its use across federal agencies, the National Security Agency is reportedly moving forward with its own deployment. According to reports from Axios, the NSA is among a select group of organizations utilizing Mythos Preview, a specialized model Anthropic describes as uniquely capable in computer security tasks.
This internal adoption comes amid a volatile legal climate. In February, President Trump ordered government agencies to cease using Anthropic’s services following a breakdown in contract negotiations regarding military safeguards. Anthropic responded with a series of lawsuits against the Department of Defense, successfully securing a preliminary injunction to temporarily block the restrictive "risk" designation. This legal friction, however, appears to have done little to dampen the intelligence community’s appetite for high-end security tools.
The strategic necessity of the technology was underscored last week when Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles. While the President claimed no knowledge of the meeting to reporters, the White House characterized the discussion as "productive and constructive." The dialogue suggests a potential rapprochement, or at least a recognition that the NSA’s defensive capabilities may now rely on the very models the executive branch has publicly sought to distance itself from.
With reporting from Engadget.
Source · Engadget
