XNews

OpenAI Pipeline Finds and Patches 948 Linux Kernel Privilege Escalation Flaws

OpenAI Pipeline Finds and Patches 948 Linux Kernel Privilege Escalation Flaws

OpenAI's security researchers built an autonomous pipeline that discovered, exploited, and patched Linux kernel vulnerabilities. In one run with an upcoming cyber model, it reproduced 1,724 vulnerabilities and generated working exploits for 948 local privilege escalation flaws at a total API cost of $1.2 million.

Event version 1.

Key insights

Pipeline covers full vulnerability lifecycle

The pipeline finds security-relevant code, scans for vulnerabilities, reproduces them in isolated environments, tests exploitability, and generates candidate patches. Human researchers review results before reporting vulnerabilities and proposing fixes to maintainers.

Run reproduced 1,724 vulnerabilities

The run reproduced 1,724 vulnerabilities and produced proofs of concept for 291 information leaks. It generated working exploits for 948 distinct local privilege escalation vulnerabilities on a stock Linux desktop configuration with standard security protections enabled.

Cost was about $1,279 per vulnerability

The full run cost $1,213,430 in API credits, about $1,279 per distinct local privilege escalation vulnerability. Applying historical Pwn2Own and TyphoonPWN bounty rates gives a hypothetical value of $18.9 million to $66.3 million.

Agents found at least 10 novel exploitation techniques

The agents discovered and used at least 10 Linux exploitation techniques that are novel to the best of the researchers' knowledge and all consulted experts. This suggests frontier LLMs can discover and patch new approaches to exploitation beyond known techniques.

Work shared with Linux maintainers

OpenAI stated its goal is to help defenders find and fix vulnerabilities before attackers can use them. The company is working with Linux maintainers and Patch the Planet partners on reports and patches.

Source posts