OpenAI's AI agent finds 948 Linux kernel privilege flaws
OpenAI said its autonomous system, built on an upcoming cyber model, identified, verified and drafted fixes for 948 local privilege-escalation flaws in the Linux kernel during an October research run. It also produced working exploits for those flaws.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
OpenAI said an autonomous pipeline built with an upcoming cyber model found, validated and wrote patches for 948 local privilege-escalation vulnerabilities in the Linux kernel. The work was done in a mid-October research run. OpenAI said the pipeline also generated working exploits for those flaws.
Why it matters
The findings show that AI systems can locate and exploit serious security weaknesses in widely used software at scale.
Timeline
- Mid-October 2026: OpenAI's prototype pipeline found and exploited 948 unique local privilege-escalation vulnerabilities in Linux during a research run.
- Nov 4, 3:00 AM ET: OpenAI announced the pipeline's results, saying it found, validated and generated patches for 948 Linux kernel vulnerabilities.
Coverage
- AI Agents Got Root on Linux via 948 Kernel Flaws in OpenAI Research (securityweek.com)
- OpenAI Just Found Nearly 1,000 Ways To Gain Root Access In Linux (hothardware.com)
- OpenAI’s agent uncovers 1K Linux kernel flaws, saving millions of dollars (cybernews.com)
- OpenAI says its AI can break into Linux at scale (coindesk.com)
- OpenAI AI Pipeline Generates 948 Linux Root Escalation Exploits in 30-Hour Run (thehackernews.com)
- OpenAI's AI pipeline turned 2,308 Linux kernel bugs into 948 root exploits (startupfortune.com)
- OpenAI Pipeline Finds 948 Linux Kernel Privilege Escalation Flaws (unite.ai)
- Dealroom.co | OpenAI's AI agents built 948 working Linux root exploits in one run, for $1.2M in API credits (app.dealroom.co)
More from Technology & AI
- China puts humanoid robots to work serving food and sorting parcels
- Newsom proposes sweeping AI safety rules for Washington
- Google releases Gemini 4 Neon AI model with million-token context
- Microsoft and Intel widen their cloud chip partnership
- Google DeepMind rolls out Genie 4 to select subscribers
- AI helped AMD shorten chip design work, OpenAI says