CISA Utilizes Anthropic's Mythos for Government Code Audits

• •CISA is using Anthropic's Mythos to scan for vulnerabilities in government software.
• •The audits have reportedly uncovered numerous security flaws, though details are scarce.
• •Anthropic's relationship with the U.S. government has been complicated by previous tensions.

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is employing Anthropic's AI model, Mythos, to audit government software for vulnerabilities. This initiative aims to identify security flaws that could be exploited by foreign intelligence or cybercriminals. Sources indicate that the audits have already revealed numerous vulnerabilities, although specifics on the nature and severity of these flaws remain undisclosed. The use of Mythos reflects a growing trend in government adoption of AI tools for cybersecurity purposes. Anthropic's relationship with the U.S. government has been complex, marked by a previous designation as a supply-chain risk by the Pentagon due to its refusal to remove certain safeguards. Despite these tensions, Mythos has been operational within the National Security Agency (NSA) since at least April. The White House's recent demands for restrictions on the public version of Mythos led to a temporary global shutdown of the model, which was lifted last week. The ongoing audits are part of a broader effort to enhance digital security across government systems.