AI Agents Got Root on Linux via 948 Kernel Flaws in OpenAI Research

An un­re­lea­sed Ope­nAI cyber model found and ex­ploi­ted 948 dis­tinct Linux ker­nel pri­vi­le­ge es­ca­la­tion bugs for about $1.2 mi­llion. The post AI Agents Got Root on Linux via 948 Ker­nel Flaws in Ope­nAI Re­search ap­pea­red first on Se­cu­rity­Week .