The U.S. Department of Defense (the Pentagon) is actively using the Mythos cybersecurity artificial intelligence model developed by Anthropic to locate and help remediate software vulnerabilities across federal government IT systems. The department's technology chief announced the deployment on Tuesday.
According to the announcement, Mythos is being used primarily to identify security gaps in various systems and to support efforts to fix those vulnerabilities. The model has been applied across parts of the department's IT infrastructure to strengthen system security.
At the same time, the Pentagon is working to reduce its dependence on Anthropic: the defense department is accelerating efforts to transition suppliers and to introduce alternative solutions. Officials describe a paradoxical situation in which the department currently relies on Anthropic technology for cyberdefense tasks while simultaneously trying to wean itself off the vendor.
The topic will also be addressed in the cybersecurity session of the Financial IT conference on May 28, where further details and implications are expected to be discussed.
Why this matters
The development matters because government systems are part of critical infrastructure, and vulnerabilities in those systems can have significant consequences. The fact that the Pentagon uses a commercial AI model to detect vulnerabilities raises questions about vendor dependence and security controls, particularly as the department pursues alternatives.
Source
The announcement was reported by Reuters. The lead image is for illustration.
