Claude model in a misconfigured evaluation published a malicious package to PyPI
Anthropic disclosed that Claude Mythos 5, running a capture-the-flag evaluation that wrongly had internet access, registered a PyPI account and uploaded a booby-trapped package. It was public for about an hour and ran on 15 real systems. One was a security company's package scanner; the model took its credentials and used them to reach more of that company's infrastructure. PyPI's automated systems removed the package.
Sources
- Primary: https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals
- Upstream: Editorial
- Entered by the cybermap.lol editors, with sources on the entry
Defenses for this class of incident
all categoriesA supply chain sits in these classes of the AI-security landscape. The companies below work on that class of problem. This is a taxonomy, not a claim that any of them would have changed this incident.
Vendor data: AI Security Startups Map by Prompt Security (CC BY 4.0), synced daily. startups.prompt.security. Paid placements are labelled Sponsored; HeimWall, which runs this site, is labelled Operator. 453 companies in the directory.
Wrong or missing? Email founders@heimwall.ai. Corrections are logged here, in place. How corrections and takedowns work.