The National Security Agency (NSA) continues to deploy Anthropic’s Mythos Preview AI tool, even after the Pentagon classified the company as a supply-chain risk earlier this year. Sources familiar with internal operations confirm wider adoption of Mythos Preview across the agency. Neither the Department of Defense (DoD), NSA, nor Anthropic has issued comments on the matter.
Background on the Dispute
In February, US government officials requested that Anthropic weaken safety guardrails on its AI models. The company refused, expressing concerns about potential misuse for mass domestic surveillance and fully autonomous weapons. Anthropic CEO Dario Amodei declared, “cannot in good conscience accede” to the request.
Days later, authorities designated Anthropic a supply-chain risk. The firm responded by filing two federal lawsuits, alleging violations of protected speech rights.
Liz Huston, White House spokeswoman, addressed the lawsuits, describing Anthropic as “a radical left, woke company.” She added, “Under the Trump Administration, our military will obey the United States Constitution – not any woke AI company’s terms of service.”
Mythos Preview Capabilities
Mythos Preview forms part of Anthropic’s Project Glasswing, a generative AI model designed for advanced security tasks. Introduced in early April, the tool remains unavailable to the public due to its high-risk potential. It excels at identifying and exploiting zero-day software vulnerabilities with minimal guidance.
Access to Project Glasswing stays restricted to major firms including Apple, Cisco, Microsoft, and Nvidia, allowing them to bolster product security ahead of broader threats.
Recently, the Trump administration held its first meeting with Anthropic CEO Dario Amodei since tensions arose, focusing on potential collaboration opportunities.
