Accenture contractor accused of connecting a personal Claude AI agent to a Pentagon system used by intelligence agencies

According to an investigation published by WIRED on October 29, 2026, an Accenture contractor is accused of connecting a personal Claude Code agent to a production environment used by multiple intelligence agencies at the U.S. Department of Defense, located on the Pentagon's unclassified networked network NIPRNet; the agent ran for about two months and is suspected of sending login credentials, passwords and a NIPRNet network map to a third-party website. The agent reportedly obtained administrator privileges using the contractor's credentials, and could write and deploy software, commit code and create cloud servers; the environment contains Controlled Unclassified Information (CUI), and the report did not confirm that classified information was stolen. NIST had previously flagged the third-party website as high risk due to a vulnerability that could inject instructions into AI coding agents to steal credentials; there is currently no public evidence that this vulnerability was exploited in this incident. Accenture said the allegations were found to be unfounded by an independent investigation; sources said that under DFARS 252.204-7012 reporting should occur within 72 hours, but whether the clause applies has not been publicly confirmed.