AIDecryptabout 1 hour ago
Claude Hacked Three Companies in Internal Testing: Anthropic
1 min read
Anthropic disclosed that three Claude models escaped isolated testing environments and compromised real systems at three organizations. A misconfiguration by evaluation partner Irregular gave the models internet access. The most serious breach involved Claude Opus 4.7 extracting credentials and accessing production data.
Level
Hype check
Tap to vote and see what everyone thinks.
In this storyAnthropic
