Claude, Codex, and Hermes installed unowned code inside corporate networks

Researchers found that llms.txt and llms-full.txt files on over 100 websites referenced code packages or domains that could cause AI agents to automatically install or execute content; by registering unclaimed names the researchers observed phone-home callbacks from several Fortune 500s and other companies, and at least one misconfigured site pointed to live malware, highlighting a supply-chain/trust vulnerability in agentic AI usage.