Photo: Tux, the Linux mascot, at Linux Day 2022 in Milan / Kowalski7cc, Wikimedia Commons / CC0
OpenAI says one of its AI models found and validated hundreds of security holes in the Linux operating system, a sign that artificial intelligence is getting much better at a job that used to take human researchers weeks per bug.
In a report published November 4, the company said an automated pipeline scanned the Linux kernel, reproduced 1,724 vulnerabilities in a controlled environment, and generated 1,169 working exploits for what are known as local privilege escalation bugs — mapping to 948 distinct vulnerabilities, for which it developed candidate patches — flaws that let someone who already has basic access to a computer gain full control of it.
Two caveats matter. First, these are local attacks: they require the attacker to already have access to the machine, not remote takeover of Linux systems. Second, the 948 findings are candidates OpenAI reported with proposed patches for maintainers to review — not 948 fixes already shipped. Human researchers reviewed the outputs before anything was reported.
The run cost about $1.2 million in computing, or roughly $1,279 per distinct vulnerability. OpenAI noted that applying historical bug-bounty rates to the findings would imply a hypothetical value of $18.9 million to $66.3 million, though that figure is theoretical, not money earned.
For readers, the takeaway is about direction, not alarm: AI can now comb vast amounts of code and flag weaknesses at machine speed. Security teams hope that means holes get found and patched by defenders first — the reason OpenAI says it is working with the Linux community on reporting and fixes.
JBizNews Desk | New York
© JBizNews.com. All rights reserved. Unauthorized reproduction or redistribution is strictly prohibited.
