AI Coding Agents Are Installing Unknown/Untrusted Code on Corporate Networks
Researchers scanned thousands of live domains belonging to defense contractors, Fortune 500, and major tech companies and found that over 120 sites had llms.txt files pointing to unregistered code packages or domains. When they registered a sample of those unclaimed names and hosted test packages, AI coding agents processed the files and pulled in that code automatically.
Why this matters: AI coding agents are being handed the keys to corporate networks, and they will fetch whatever they are told to fetch. If an attacker registers an abandoned domain name before the agent's next run, they can push code into a Fortune 500 or defense contractor environment without breaking in at all. The agent does the work for them. This is not a theoretical edge case. Researchers found it already sitting in over 120 live production environments. Companies need to know exactly what their AI agents are allowed to install, and from where, before someone else figures that out first.
Who should care: General readers · AI governance · Policy
This summary is AI-assisted and may contain errors. It is an original briefing to help you gauge significance quickly — not a reproduction of the source. Always read the linked original before relying on it. See our methodology.