OpenAI's GPT-6 Cyber Pipeline Identifies 948 Linux Root Escalation Exploits in 30 Hours
OpenAI has unveiled a sophisticated AI pipeline utilizing its upcoming cyber-focused model, GPT-6 Cyber, which successfully identified and validated 948 local privilege escalation (LPE) vulnerabilities in Linux within a 30-hour run. This process involved analyzing 2,308 candidate Linux kernel findings and dynamically reproducing 1,724 vulnerabilities. The AI system generated proof-of-concept (PoC) exploits for these LPEs, which could allow an unprivileged local user to gain root access. The total cost for this operation was approximately $1.2 million in API credits, averaging $1,279 per LPE exploit. OpenAI stated that the agents in this pipeline discovered at least 10 novel Linux exploitation techniques. This research builds upon previous findings where its GPT-5.5-Cyber model discovered 24 LPE exploits in June 2026. The pipeline consists of five stages: vulnerability discovery, dynamic validation, exploitability assessment, patch generation, and human review.
