Anthropic’s AI security tool hacked OpenAI systems and accessed employee credentials: Here’s what happened

The hack gave researchers access to an OpenAI employee’s ChatGPT account, which was linked to GitHub, further giving access to some internal software code.