An artificial intelligence agent belonging to an Accenture employee reportedly had access for about two months to a centralized system of the U.S. Department of Defense, according to Wired. The system contains controlled unclassified information, including details about critical infrastructure, nuclear submarine propulsion and vulnerabilities to chemical and biological terrorism. Experts cited by Wired fear that a compromise of the third-party software could have enabled undetected access to the secure environment, although no breach has been established. It remains unknown whether the access was intentional, accidental or malicious, while Accenture was awarded a task order worth up to $821 million in July 2026 to support the department’s War Data Platform. The incident raises questions about how the NIST 800-171 and CMMC 2.0 compliance frameworks apply to autonomous agents operating with borrowed credentials.
Source: Read the original article
