NSA Deploys Anthropic's Controversial Mythos AI Model Despite Pentagon's Security Risk Label
The United States National Security Agency (NSA) is reportedly utilizing Anthropic's Mythos Preview, the company's most controversial artificial intelligence model to date. This development occurs despite the US Department of War, which oversees the NSA, having designated Anthropic as a 'national security risk' in March of this year.
Widespread Usage Within Government Agencies
An Axios report, citing two individuals familiar with the situation, indicates that the Mythos model is being employed more extensively within the department. While the specific applications by the NSA remain unclear, the report notes that other organizations with access to the model are primarily using it to scan their own environments for exploitable security vulnerabilities.
Anthropic's Public Feud with the Pentagon
The US Department of War classified Anthropic as "a supply chain risk," a designation historically reserved for foreign companies. The conflict originated when Anthropic declined a Pentagon ultimatum demanding "full, unrestricted access" to its AI tool Claude, citing ethical apprehensions related to mass surveillance and fully autonomous weapons systems.
In a sharply critical Truth Social post at the time, President Donald Trump characterized the company's leadership as "Leftwing nut jobs," signaling a deepening rift between the White House and key AI suppliers to the Pentagon. Trump asserted, "THE UNITED STATES OF AMERICA WILL NEVER ALLOW A RADICAL LEFT, WOKE COMPANY TO DICTATE HOW OUR GREAT MILITARY FIGHTS AND WINS WARS!"
Anthropic contested the ban and initiated legal proceedings. On March 26, US District Judge Rita Lin granted the AI company's request for a preliminary injunction, suspending both the Presidential Directive ordering federal agencies to cease using Anthropic's technology and Defense Secretary Pete Hegseth's designation of the company as a "supply chain risk."
Anthropic's Decision to Withhold Mythos from Public Release
Upon announcing the Mythos AI model, Anthropic declared that it would not publicly release its latest AI model, Claude Mythos Preview, due to concerns that it could destabilize the cybersecurity landscape. In a blog post, the company described Mythos as capable of autonomously finding, analyzing, and exploring software vulnerabilities at scale, in some cases more effectively than human experts.
Labeling it a "watershed moment," Anthropic cautioned that even non-specialists could potentially use Mythos to uncover and exploit sophisticated flaws. During testing, Mythos reportedly detected thousands of critical vulnerabilities, including zero-day flaws that typically require elite human teams months to identify.
According to a Business Insider report, cybersecurity specialists warn that if Mythos were made publicly available, malicious actors could benefit first by generating phishing campaigns, deepfakes, or exploit chains instantaneously. The model's advanced capabilities present both significant opportunities for defense and substantial risks if misused.
