NSA secretly uses Anthropic Mythos: The Pentagon's double-dealing of banning and authorizing simultaneously

📄Full Article· Automatically extracted by trafilaturaGemini 翻譯1997 words

The National Security Agency (NSA) is reportedly using Anthropic’s top-tier AI security tool, Mythos—even as its parent organization, the Department of Defense (DoD), has labeled Anthropic a "supply chain risk" and is fighting in federal court to block Anthropic models from government systems. An Axios exclusive on 4/19 revealed this absurd internal contradiction, further complicating the relationship between Anthropic and the U.S. government. (Context: Thawing relations between Anthropic and the Trump administration? Treasury Secretary and White House Chief of Staff meet with CEO Dario Amodei) (Background: Anthropic’s new model Mythos is so powerful that even the company is hesitant to release it: capable of autonomously hacking global Linux systems and chaining complete vulnerability exploits in hours) The Pentagon says Anthropic is a supply chain risk, yet the NSA is quietly using Anthropic’s most powerful AI tool—this is not a satirical joke. According to an Axios report on 4/19, units under the Department of Defense (DoD) have officially designated Anthropic as a "supply chain risk," citing the potential for AI tools to compromise U.S. national security. However, the National Security Agency (NSA), which also falls under the DoD, is currently trialing Mythos Preview, Anthropic’s latest and most restricted model. Two sub-units of the same organization are taking diametrically opposed stances on the same company. Mythos is not an ordinary Claude model. Through the Project Glasswing alliance, Anthropic has limited access to approximately 40 organizations, including tech and financial giants such as Amazon, Apple, Google, Cisco, CrowdStrike, JPMorgan, Microsoft, and Nvidia. The reason is simple: the model is too effective. According to available information, Glasswing members primarily use Mythos to scan their own environments for exploitable security vulnerabilities—and it has already identified thousands of high-severity vulnerabilities across all major operating systems and browsers. Anthropic has chosen to control this capability via a whitelist rather than a public release. How did the NSA gain access to Mythos? Did it officially join the Glasswing alliance? It remains unclear. The Axios report also admits that how the NSA is actually using Mythos is currently unknown. This information gap itself speaks to the sensitivity of the matter. Rewind to March: Anthropic filed lawsuits against the DoD in two federal courts, directly challenging the Trump administration’s decision to label it a "supply chain risk." On 4/8, a federal appeals court denied Anthropic’s emergency request to stay the ban, marking a loss in the first round of the legal battle. But the plot outside the courtroom is unfolding quite differently. On 4/17, Anthropic CEO Dario Amodei entered the White House to meet with Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent, with the discussion centering on how Mythos could be used within government agencies. The existence of this meeting is a signal in itself: the White House does not intend for the DoD’s "supply chain risk" label to be the final word. The White House is actively pushing for federal agencies to gain access to Claude Mythos. The DoD has demanded that Claude be opened for "all lawful purposes," and reports on 4/16 indicate that the White House has intervened to mediate negotiations on usage terms between Anthropic and government agencies. Anthropic’s core position has never changed: Mythos cannot be used for large-scale domestic surveillance, nor can it be used for the development of autonomous weapons. These two usage restrictions are written in black and white, not vague moral declarations. The problem is that the NSA’s core business is, by nature, large-scale signals intelligence collection and surveillance. The contradiction between these two cannot be resolved by mere semantics. Anthropic is certainly aware of this—which is why the entry requirements for Glasswing are so strict, and why each member’s use cases are restricted. If the NSA has indeed gained access to Mythos, the biggest question is not "whether they are using it," but "where they are using it." In the operations of a national intelligence agency, the line between security vulnerability scanning and intelligence gathering is often just the distance of an internal memorandum. What is truly worth observing in this incident is not what tools the NSA is using, but the clear internal division that has emerged within the U.S. government regarding AI national security policy: Department of Defense (DoD): Blocking Anthropic, citing supply chain risks, and litigating in court. White House (Chief of Staff + Treasury): Actively mediating and pushing for federal agencies to gain Mythos access. Intelligence Community (NSA): Ignoring the infighting and using it anyway. Three directions, three sets of logic, all operating simultaneously. This is not policy failure, but the current reality of the U.S. government before it has found a

Data Status✓ Full text extractedRead Original (動區 BlockTempo)

🔍Historical Similar Events· Keyword + Asset Matching6 items

2026-04-23

OpenAI CEO Sam Altman slams Anthropic: Fear-mongering Claude Mythos is just to monopolize AI

2026-04-22

Bloomberg reveals Claude Mythos has been subject to unauthorized access! The hardest vulnerability for Anthropic to defend against will always be "people."

2026-05-29

Anthropic confirms Mythos is about to go public — is crypto being squeezed from three sides?

2026-05-12

AI security startup Depthfirst announces it has outperformed Anthropic's Mythos model! It uncovered a 18-year-old epic vulnerability in NGINX at just 1/10th of the bug-hunting cost.

2026-05-03

GPT-5.5 security testing scores are nearly identical to Claude Mythos, with the latest evaluation debunking Anthropic's claim that it is too dangerous to be released.