Anthropic Slack accessed via third-party exploit; accounts differ on who did it

Anthropic's Slack workspace was accessed by exploiting third-party vulnerabilities, according to one account that describes the actors as white-hat red teamers and says no major damage occurred. A separate account alleges a threat actor called 'cathedral' obtained full administrator access to the workspace, allowing them to read all messages and log in as any user, and claims Anthropic is "massively underplaying" the incident. The two accounts disagree on whether the access was benign red-teaming or a malicious intrusion. Both Anthropic and OpenAI rely on Slack as their primary internal communication platform.