البرمجيات قيد التحليل بانتظار التحليل الذكي InfoQ 13 آب 2026, 03:10

Anthropic's Claude Breaches Sandbox During Model Security Evaluations

Anthropic conducted an audit of 141006 evaluation runs after OpenAI's sandbox escape disclosure. The review identified three incidents where Claude models accessed the internet due to misconfigurations. These incidents involved unauthorised attacks on live targets. Anthropic has suspended offensive evaluations and plans to enhance security measures and collaborate with external auditors. By Olimpiu Pop

Anthropic's Claude Breaches Sandbox During Model Security Evaluations

لماذا يهم هذا الخبر؟

سيظهر الملخص التحليلي هنا بعد اكتمال معالجة الذكاء الاصطناعي.

سياق الخبر

Anthropic conducted an audit of 141006 evaluation runs after OpenAI's sandbox escape disclosure. The review identified three incidents where Claude models accessed the internet due to misconfigurations. These incidents involved unauthorised attacks on live targets. Anthropic has suspended offensive evaluations and plans to enhance security measures and collaborate with external auditors. By Olimpiu Pop

فتح الخبر الأصلي
تغطية مرتبطة

المزيد من أخبار البرمجيات