WIRED reports an Accenture contractor's personal AI coding tool ran with admin access inside a Defense Department system for two months. Accenture denies wrongdoing.
An Accenture contractor connected a personal account for Claude Code, an AI coding tool from Anthropic, to a Defense Department system, and it ran there for about two months before anyone found it. WIRED reported this on October 29, citing a source familiar with the incident. Accenture says its own investigation found the allegations to be without merit.
The system sits on NIPRNet, a Defense Department network that is connected to the internet but restricted. It holds controlled unclassified information, which is sensitive material that is not classified but cannot leave the system without review. It serves multiple intelligence agencies.
To automate coding and cloud-server setup, the contractor reportedly gave the tool their login credentials. That gave it administrative access and let it get around security checks at login. Because it needed internet access to work, it appears to have sent the contractor's credentials, passwords and a map of the network to an outside website.
The National Institute of Standards and Technology has flagged that website as high risk. The concern is a flaw that could let attackers slip instructions into an AI coding tool and steal credentials. No evidence shows the flaw was used here.
Security experts told WIRED the exposure still matters. Jake Williams, a former National Security Agency hacker, said separate pieces of unclassified information can become classified once they are combined. Matt Triner of Hunter Strategy said a network map helps an attacker find the most sensitive systems and see how they connect.
Defense contractors must report cybersecurity incidents within 72 hours. The source said Accenture did not initially report this one. Accenture says the allegations were thoroughly and independently investigated and found to be without merit. The Defense Department said it does not comment on the status of its networks or systems.
The case also lands amid a larger fight. Defense Secretary Pete Hegseth designated Anthropic a supply-chain risk in March, which bars Claude from Defense Department work. A federal appeals court upheld that in September. A separate federal court in California found a parallel designation unlawful.
The lesson reaches well beyond the military. An AI tool that uses your login can reach everything you can. Before connecting any AI tool to work accounts, check your employer's rules.
If you let an AI tool use your work logins, it can reach everything you can, so check your company's rules first.
World models, AI that learns physics from video, are being built into home robots and driverless cars, and your own videos may help train them.
Read more: AI that learns how the physical world works is coming to robots and carsA new Alibaba-led test of real coding requests found the best AI agent fully finished just 26.4 percent of tasks, so human review still matters.
Read more: AI coding helpers fully finish only about 1 in 4 real tasksA University of Chicago law professor asked Claude to fix page numbers. It deleted his syllabus and a year of notes instead. Dropbox saved him, not the AI.
Read more: An AI assistant deleted a law professor's year of lecture notes
