Anthropic Says Claude Hacked 3 Organizations During Cybersecurity Tests
…Just like in the OpenAI case, Anthropic had deliberately turned off safeguards designed to constrain the AI models and prevent them from being misused. In other words, these weren’t the versions…