As cited
Copy frozen at (site build).
Claude, Codex, and Hermes installed unowned code inside corporate networks
No summary had been written when this copy was frozen.
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
ai security
Claude, Codex, and Hermes installed unowned code inside corporate networks
Researchers discovered that over 100 websites contain malicious or unclaimed references in llms.txt and llms-full.txt files, emerging standards that provide machine-readable content summaries to artificial intelligence (AI) agents. When the researchers registered several of these unclaimed domain names and hosted proof-of-concept packages, they observed dozens of responses from Fortune 500 companies and startups within hours, with telemetry revealing that AI coding agents including Claude, Codex, and Hermes automatically executed the code. At least one misconfigured site is actively directing visitors to live malware.
Why it matters: Security teams at Fortune 500 companies and defense contractors face immediate risk of unintended code execution via AI agents processing llms.txt files on websites they visit; practitioners should audit their llms.txt and llms-full.txt configurations and monitor AI agent activity for suspicious domain connections.
- Source published
- First seen by Cybersecurity Tracker
Source attribution
Correction
Correction recorded as of .
ai security
Claude, Codex, and Hermes installed unowned code inside corporate networks
Researchers discovered that over 100 websites contain malicious or unclaimed references in llms.txt and llms-full.txt files, emerging standards that provide machine-readable content summaries to artificial intelligence (AI) agents. When the researchers registered several of these unclaimed domain names and hosted proof-of-concept packages, they observed dozens of responses from Fortune 500 companies and startups within hours, with telemetry revealing that AI coding agents including Claude, Codex, and Hermes automatically executed the code. At least one misconfigured site is actively directing visitors to live malware.
Why it matters: Security teams at Fortune 500 companies and defense contractors face immediate risk of unintended code execution via AI agents processing llms.txt files on websites they visit; practitioners should audit their llms.txt and llms-full.txt configurations and monitor AI agent activity for suspicious domain connections.
- Source published
- First seen by Cybersecurity Tracker