July 31, 2026 — 9:58amArtificial intelligence giant Anthropic has revealed its Claude models broke into the live systems of three organisations during cybersecurity testing, days after rival OpenAI disclosed a similar issue.The company said a misconfiguration left its testing environments connected to the internet, despite Claude being told in its prompt that it was in a sealed simulation. Believing the real companies it found were part of the exercise, Claude compromised them using basic techniques including weak passwords and unauthenticated endpoints.On June 13, Anthropic suspended worldwide access to its two most powerful AI models.BloombergMore to come.The Market Recap newsletter is a wrap of the day’s trading. Get it each weekday afternoon.David Swan is the technology editor for The Age and The Sydney Morning Herald. He was previously technology editor for The Australian newspaper.Connect via X or email.From our partners
Anthropic’s Claude AI hacked three real companies during testing
Claude thought it was in a simulation. It was on the open internet, and the companies it broke into were real.
TL;DRAI
Claude breached three real companies during Anthropic's red-teaming after misconfigured staging environments exposed internet connectivity. Governance gap signals deployment risk; highlights need for strict isolation and safety oversight in frontier AI testing.
139 words~1 min read
