Despite the Pentagon’s classification of Anthropic as a “supply chain risk,” the National Security Agency has begun utilizing the company’s restricted Mythos Preview AI model, revealing internal conflicts within the defense community regarding advanced AI technologies.
This development indicates that the government’s growing appetite for state-of-the-art AI capabilities may supersede internal institutional disagreements, potentially strengthening Anthropic’s standing in the profitable federal marketplace.
Key Takeaways
- NSA deploys Mythos despite Pentagon blacklist of Anthropic
- Model used primarily for cybersecurity vulnerability scanning
- Trump administration-Anthropic relations appear to be thawing
Conflicting Government Positions
The intelligence agency’s deployment of Mythos Preview follows the Department of Defense’s February decision to classify Anthropic as a supply chain risk, occurring just weeks later 1. The Pentagon severed ties with the AI company after Anthropic declined to permit unrestricted model access for widespread domestic surveillance and autonomous weapons programs.
Earlier this month, Anthropic unveiled Mythos as an advanced cybersecurity model while declining public distribution due to concerns about its offensive cyber capabilities 2. The company restricted distribution to roughly 40 organizations, publicly identifying only twelve recipients through its Project Glasswing program.
Cybersecurity Applications
Intelligence sources reveal the NSA employs Mythos principally for environmental vulnerability scanning, mirroring deployment patterns observed in other organizations with model access 3. The UK’s AI Security Institute has similarly confirmed Mythos access, indicating international intelligence collaboration on AI-enhanced cybersecurity platforms.
Anthropic describes the model as their “most capable yet for coding and agentic tasks,” providing potentially unmatched abilities to detect cybersecurity vulnerabilities and develop exploitation strategies 4. These capabilities make it exceptionally valuable for defensive cybersecurity initiatives.
Improving Relations
The NSA’s implementation coincides with improving ties between Anthropic and the current administration. Last Friday, CEO Dario Amodei conducted meetings with White House Chief of Staff Susie Wiles and Treasury Secretary Scott Bessent regarding Mythos implementation across government agencies.
Administration officials characterized the meeting as “productive,” with subsequent phases expected to address how departments outside the Pentagon will engage with the model 5. This development suggests the administration may be working to resolve the institutional discord that has complicated AI acquisition processes.
Market Implications
This situation underscores the strategic value of AI companies in national security contexts, potentially offering Anthropic a competitive edge in government contracting. Although the company confronts legal challenges from the Pentagon’s supply chain classification, ongoing intelligence agency usage may demonstrate the operational necessity of its technology.
The circumstances reflect broader AI governance challenges, where security agencies’ operational requirements may clash with wider policy limitations. For investors monitoring AI adoption trends, government applications frequently precede commercial rollouts and indicate technological readiness.
Not investment advice. For informational purposes only.
References
1Rebecca Bellan (April 20, 2026). “NSA spies are reportedly using Anthropic’s Mythos, despite Pentagon feud”. TechCrunch. Retrieved April 20, 2026.
2Maria Curi and Sam Sabin (April 19, 2026). “Scoop: NSA using Anthropic’s Mythos despite blacklist”. Axios. Retrieved April 20, 2026.
3Reuters (April 19, 2026). “US security agency is using Anthropic’s Mythos despite blacklist, Axios reports”. Reuters. Retrieved April 20, 2026.
4Rob Williams (April 19, 2026). “NSA is said to use Anthropic’s new AI model despite Pentagon dispute”. Seeking Alpha. Retrieved April 20, 2026.
5Nick Wingfield (April 20, 2026). “NSA Uses Anthropic’s Mythos Despite Pentagon Blacklist”. The Information. Retrieved April 20, 2026.