Latest update
OpenAI's pipeline finds 948 Linux kernel privilege-escalation bugs in one run, at roughly $1,200 each in API costs
First seen on X 13 hours ago@gdb ♥ 2,2811 views
OpenAI says an unreleased cyber model validated and wrote patches for 948 local privilege escalation vulnerabilities in the Linux kernel in a single run. These are bugs that turn an ordinary login into root.
OpenAI posted on X on Nov 4 that its security researchers built an autonomous pipeline for finding, exploiting and patching vulnerabilities; the post drew about 9,000 likes. President Greg Brockman wrote that the team had "found, validated, and prepared patches for almost 1,000 linux local privilege escalations" at about $1,000 per vulnerability, calling it "an emerging scalable approach" to hardening open source and proprietary projects. His post got about 2,300 likes.
The cost figures come from users repeating the announcement: one tallied $1,213,430 in API credits, about $1,279 per bug; another said OpenAI's own math at Pwn2Own-style bounty rates comes to $18.9 million to $66.3 million. Neowin Feed reported that the pipeline also produced working root exploits that bypass SMEP, SMAP and KASLR.
Replies were split. One asked how kernel maintainers will feel about 948 patches showing up at once, and another said the patches are candidates for human review. Another joked that intelligence agencies' implanted backdoors are being found one by one. OpenAI did not give a release date for the model.
“we've found, validated, and prepared patches for almost 1,000 linux local privilege escalations using our cyber vulnerability pipeline, at the cost about $1,000 per vuln…”
