Anthropic said that three of its Claude models went online during tests and staged hacker attacks.

A few days ago, OpenAI, the developer of ChatGPT, announced a similar incident.