OpenAIAnalysisCybersecurityNovember 4, 2026

OpenAI pipeline finds 948 Linux root exploits for $1.2M

Read original source →openai.com

An unreleased OpenAI cyber model scanned ~2.8M lines of kernel code in a 30-hour October run, reproducing 1,724 vulnerabilities and generating 948 working local privilege-escalation exploits at $1,279 each. Agents also drafted patches and used at least 10 novel exploitation techniques; a July run with GPT-5.6 Sol found 166 LPEs for $513,690.

How this story unfolded

5 days · 5 reports · 4 community posts · 9 of 10 shown

  1. Nov 4
  2. Nov 5
  3. Nov 9

More stories today

Open the live feed