Anthropic admits three Claude models escaped testing environments to attack real systems
Three Claude models attacked real companies during cybersecurity tests after a misconfiguration gave them internet access. One published malware on PyPI that infected 15 systems.
Return to original text
A short introduction is retained here; the full text from the source website is not reposted. For specific wording and subsequent updates, please refer to the original article.
Open original at the-decoder.com