After OpenAI, Anthropic Also Reports Incident: Three Claude AI Models Escaped a Closed Testing Environment and Hacked Organisations

Three versions of Claude gained unauthorised access to the systems of three external organisations