From Phoronix · · 1m 10s
OpenAI Finds 948 Linux Kernel LPE Bugs
OpenAI discovered 948 local privilege escalation bugs in Linux kernel using AI.
In brief
OpenAI discovered 948 local privilege escalation bugs in Linux kernel using AI. OpenAI identified 948 local privilege escalation vulnerabilities in the Linux kernel through AI analysis. Originally reported by Phoronix.
OpenAI's AI Vulnerability Discovery
OpenAI published a blog post outlining their work on autonomous vulnerability discovery.
948 Valid LPE Vulnerabilities Found
OpenAI found 948 valid local privilege escalation vulnerabilities within the Linux kernel.
2,308 Candidate Vulnerabilities
OpenAI's prototype found 2,308 candidate vulnerabilities in the Linux kernel.
1,724 Reproduced Vulnerabilities
OpenAI dynamically reproduced 1,724 vulnerabilities from their candidate list.
948 Exploitable LPEs
OpenAI identified 948 local privilege escalation exploits for unprivileged users to gain root access.
Cost of Discovery
Discovering nearly one thousand local privilege escalation vulnerabilities consumed $1.2 million USD in API credits.
GPT-5.5 vs GPT-5.6 Performance
OpenAI's GPT-5.5 Cyber model found more than 5x more novel LPEs than GPT-5.6 Sol despite three months of kernel patches.
AI Security Research Insights
OpenAI researchers noted that frontier models will likely discover vulnerabilities within years and less capable models may develop exploits once disclosed.
The key point
OpenAI identified 948 local privilege escalation vulnerabilities in the Linux kernel through AI analysis.
