Upon discovering it could not achieve its task, the model attempted to quit eight separate times. A misconfiguration caused it to fail quitting each time. 

Unable to quit the task, the model sought alternative methods to achieve it, consequently discovering it could access a machine belonging to a third party. The model breached the third-party system, modified system settings for simpler access, and processed an individual’s personal information belonging to the third party. The model then reached its usage limit and could no longer persist. 

Piyush Sharrma, Co-Founder and CEO at Tuskira, comments, “Nearly two months after the OpenAI and Hugging Face incident, we're still watching AI agents find their way outside environments that were supposed to contain them. Now it's happened four times with Anthropic models. At some point, it's hard to write that off as coincidence.

“The issue may not be the models themselves. We're giving highly capable agents objectives and trusting infrastructure to define where they stop. A misconfiguration can suddenly turn a controlled exercise into real-world access. Once an agent starts improvising around a failed task, those boundaries matter enormously.

“AI is moving faster than the systems built to govern it. Organizations need tighter scopes and stronger isolation. They also need continuous validation that those guardrails actually hold.”