CISA adopts Anthropic Mythos AI to audit government code

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) is now using Anthropic's advanced Mythos artificial intelligence model to scan government software for security vulnerabilities, according to multiple sources familiar with the matter. The deployment marks a significant operational shift for the federal government, integrating a frontier AI system directly into its cybersecurity defense strategy.

The initiative is part of CISAs Attack Surface Evaluation work, where a specialized team uses the model to audit code repositories across government systems. The goal is to identify potential weaknesses that could be exploited by cybercriminals or foreign intelligence agencies. Sources indicate the AI-assisted audits have already uncovered a large number of vulnerabilities, though the exact scale, nature, and severity of the discovered flaws have not been publicly detailed.

This move represents a notable evolution in how government agencies approach vulnerability discovery, moving beyond reliance solely on human review and conventional tools. The integration of AI aims to surface risks earlier and more efficiently than traditional methods. Preliminary testing by other...