OpenAI AI Agents Find 948 Linux Kernel Exploits
Original title: AI Agents Got Root on Linux via 948 Kernel Flaws in OpenAI Research
OpenAI's unreleased cyber model agents found 948 unique Linux kernel LPE exploits in a single run at $1.2 million in API costs.
OpenAI used an autonomous AI pipeline powered by an unreleased cyber model to discover and exploit 948 unique Linux kernel vulnerabilities during one scan of the master branch. The agents narrowed down to low-privileged reachable areas, flagged 2308 potential bugs, confirmed 1724, and produced 1169 working local privilege escalation exploits plus 291 information leak PoCs. All testing occurred on virtual machines with protections like KASLR, SMEP and SMAP enabled. OpenAI submitted patches to maintainers and noted over 100 bugs were already patched during the scan.
- 948 unique exploits cost $1,213,430 or roughly $1,279 each
- More than five times previous GPT-5.6 Sol scan results
- Full technical details kept private for safety reasons
- Agents identified at least 10 novel Linux exploitation techniques
- OpenAI is raising Security Bug Bounty payouts
- Vendors
- OpenAI
- Products
- Linux kernel
- AI models
- GPT-5.6 Sol
From the original article
3 min read at securityweek.com
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.
By Eduard Kovacs · securityweek.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.