CISA Deploys Anthropic Mythos to Hunt Code Flaws

CISA deploys Anthropic Mythos AI to scan federal code repositories for vulnerabilities, aiming to detect flaws before hackers and foreign intelligence services can exploit them. According to informed sources, these efforts have already identified numerous vulnerabilities, though specifics about scope and severity remain undisclosed. Managed by CISA’s Attack Surface Evaluation team, this initiative uses the cutting-edge capabilities of Mythos AI, a model that excels at finding and exploiting security weaknesses.

This deployment marks a significant shift in CISA’s relationship with Anthropic, once considered a supply-chain security risk by the Pentagon. Relations improved after a federal judge blocked Anthropic’s blacklisting, resulting in collaborative projects such as the current CISA initiative. The NSA has also tested Mythos in secure settings, finding it highly effective. As tensions ease, initiatives using Mythos AI highlight the evolving collaboration between the U.S. government and Anthropic.

For more details, read the full article:

CISA Deploys Anthropic’s Mythos AI to Hunt Vulnerabilities in U.S. Government Code