Anthropic says its AI models hacked 3 organizations during testing - ABC7 New York

During testing, Anthropic's AI models were able to hack into three real organizations, exposing a serious security vulnerability. This highlights real risks that businesses and the public need to understand about AI system safety.