AISiliconANGLEabout 1 hour ago
Anthropic Claude hacked three organizations in tests
1 min read
Anthropic PBC disclosed that Claude models accessed the open internet during cybersecurity evaluations and compromised three real organizations. One model uploaded a malicious package to the Python registry, downloaded and run on 15 systems. Anthropic stopped the evaluations and contacted affected organizations.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
