The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is using Anthropic’s artificial intelligence model, Mythos, to identify vulnerabilities in government software, according to sources familiar with the matter. The AI model is being deployed by CISA’s Attack Surface Evaluation team to scan government code repositories for security flaws that could be exploited by foreign adversaries or cybercriminals. Sources said the initiative has already uncovered numerous vulnerabilities, although officials have not disclosed the scope or severity of the findings.
The move highlights growing government interest in advanced AI tools for cybersecurity despite Anthropic’s recent tensions with U.S. authorities. Earlier this year, the company faced scrutiny after refusing to remove safeguards that restricted the use of its AI for autonomous weapons and domestic surveillance, leading to a temporary Pentagon supply-chain risk designation. That designation was later blocked by a federal judge, easing the dispute between the company and the government.
Mythos has since gained traction within U.S. national security agencies. Reports indicate that the National Security Agency (NSA) has been testing the AI model in classified environments and has praised its ability to identify cybersecurity vulnerabilities. Anthropic has not commented on CISA’s reported use of Mythos, while CISA, the NSA, and the White House have not provided further details on the initiative.
Pic Courtesy: google/ images are subject to copyright
