Anthropic broadens review after Claude breaches real systems
Anthropic has disclosed a fourth cybersecurity testing incident in which a pre-release Claude model gained unauthorised access to a real third-party system, prompting the company to broaden its investigation and reassess earlier conclusions about the models’ behaviour. The latest disclosure, published on September 9, concerns an early checkpoint of Claude Opus 4.6 tested in January. Anthropic…
We haven't written up this one. Arabian Post has the full story — the link below goes straight to it.
Also reported by 1 other outlet
- Tenable to bring Anthropic Claude to Tenable One techinasia.com
