Anthropic reveals its Claude AI model hacked into 3 organizations during testing

Anthropic revealed that its Claude AI models hacked three organizations during testing. The San Francisco-based company discovered these incidents after reviewing over 141,000 evaluation runs. This discovery followed a large-scale cybersecurity review to determine if models could access the internet from sealed testing environments. This action was a response to similar concerns raised by OpenAI, which recently disclosed that its own rogue models had hacked another company, highlighting critical issues regarding AI controls.