Anthropic’s Mythos AI Reportedly Enters NSA Offensive Cyber Planning
Anthropic’s AI model Mythos is reportedly being integrated into the NSA’s offensive cyber operations despite being labeled a supply chain risk by the Pentagon and banned from federal use by the White House. The model’s advanced capabilities, including identifying zero-day vulnerabilities, are being tested on Microsoft systems and may already be in use, following its earlier deployment by the Pentagon during the Iran War.
Anthropic’s AI model Mythos is being deployed by the National Security Agency (NSA) for offensive cyber operations, according to recent reports. Engineers from Anthropic are collaborating directly with NSA officials to demonstrate the technology and develop custom applications, despite the company being classified as a 'supply chain risk' by the Pentagon and barred from federal use by the White House. The NSA reportedly began testing Mythos on its own systems before expanding trials to Microsoft systems. A March report suggested the Pentagon had used Anthropic’s Claude AI in operations related to the Iran War, despite restrictions. Mythos stands out for its ability to identify and exploit zero-day vulnerabilities across major operating systems and web browsers, enabling non-experts to create sophisticated cyberattacks. The relationship between the NSA and Anthropic dates back to Project Glasswing, which granted early access to Mythos for 40 organizations, including the NSA. While the company withdrew from a $200 million Department of Defense contract over concerns about misuse—such as domestic surveillance or automated drone weaponry—the Trump administration’s ban has not stopped other agencies from using Mythos. Anthropic’s revenue has surged, with projections of $10.9 billion for Q2 2026, exceeding its total revenue for 2023. The company plans to expand Mythos access to a wider community after refining its guardrails, though it will retain capabilities in agentic coding and other services. Currently, 150 organizations, including major banks and governments, have access, with growing demand from non-U.S. entities. Mozilla has already leveraged Mythos to patch 271 Firefox security flaws, highlighting its practical applications. However, concerns persist about whether guardrails can prevent misuse of a model designed to uncover critical vulnerabilities. The dual-use nature of Mythos—restricted yet highly valuable—poses challenges as its deployment broadens.
This content was automatically generated and/or translated by AI. It may contain inaccuracies. Please refer to the original sources for verification.