OpenAI pipeline identifies and patches 948 Linux security vulnerabilities

OpenAI has developed a prototype cybersecurity pipeline using an unreleased AI model to automatically detect and patch Linux security flaws.

The pipeline successfully validated and patched 948 local privilege escalation exploits, costing approximately $1.2 million in API credits.

This tool underscores the dual-use nature of AI in cybersecurity, as these models can both facilitate rapid exploitation and accelerate defensive patching.