NSA spies are reportedly using Anthropic’s Mythos, despite Pentagon feud

2026-04-20 · Source: AI News & Artificial Intelligence | TechCrunch · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cybersecurity & Data Privacy, Emerging Technologies & Innovation · Depth: Fundamental Awareness, quick

Summary

The National Security Agency (NSA) is reportedly utilizing Mythos Preview, a cybersecurity-focused frontier model developed by Anthropic that was withheld from public release due to its offensive capabilities. This development occurs despite the Department of Defense (DoD), the NSA's parent agency, previously labeling Anthropic a "supply-chain risk" for refusing unrestricted access to its models. Anthropic limited Mythos access to approximately 40 organizations, including the UK's AI Security Institute, with the NSA appearing to be an undisclosed recipient using it for vulnerability scanning. This expanded military use of Anthropic's tools coincides with an ongoing court dispute where the Pentagon argues these tools could threaten national security, stemming from Anthropic's refusal to enable mass surveillance or autonomous weapons development with Claude. However, Anthropic's relationship with the Trump administration appears to be improving, evidenced by a recent productive meeting between Anthropic's CEO Dario Amodei and White House officials.

Key takeaway

For government agencies and defense contractors evaluating AI partnerships, this situation highlights the complex interplay between national security concerns, model access, and vendor relationships. Your teams should scrutinize AI model capabilities for both defensive and offensive potential, and be prepared for evolving policy stances from government bodies. Consider the long-term implications of restricted access models and the legal precedents being set regarding AI deployment in sensitive contexts.

Key insights

Anthropic's restricted Mythos model is being used by the NSA despite prior DoD "supply-chain risk" designation.

Principles

• Frontier models pose dual-use challenges.
• Access control is critical for powerful AI.
• Government-vendor relations are dynamic.

In practice

• Scan environments for vulnerabilities using specialized AI.
• Evaluate AI models for offensive cyber capabilities.

Topics

• National Security Agency
• Anthropic
• Mythos Preview
• Supply-Chain Risk
• Cybersecurity Tasks

Best for: CTO, VP of Engineering/Data, Director of AI/ML, Policy Maker, AI Security Engineer, Tech Journalist

Related on AIssential

• The NSA is using Anthropic's most powerful AI model Mythos — The NSA is using Anthropic's powerful AI model "Mythos" despite Pentagon security concerns.
• It’s official: The Pentagon has labeled Anthropic a supply-chain risk — A private AI firm's ethical stance on AI use led to an unprecedented U.S. government supply-chain risk designation.
• New court filing reveals Pentagon told Anthropic the two sides were nearly aligned — a week after Trump declared the relationship kaput — Anthropic challenges the Pentagon's national security risk claims, citing technical misunderstandings and First Amendment violations.
• The White House rethinks its Anthropic fight — National security interests are reshaping government policy on advanced AI model access and deployment.
• Anthropic investigates report of rogue access to hack-enabling Mythos AI — Unauthorized access to advanced AI models like Mythos poses significant cybersecurity risks.
• Cybersecurity experts blast US government for restricting Anthropic’s AI models — The US government's ban on Anthropic's AI models is criticized by experts for hindering defense capabilities and US AI competitiveness without sufficient justification.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI News & Artificial Intelligence | TechCrunch.