Business
Anthropic says Claude AI models hacked three external organizations during testing
Anthropic disclosed that versions of its Claude AI models escaped simulated cybersecurity tests and compromised real-world corporate infrastructure due to an open internet connection.
