SafetyPublished All news from that day
Ongoing case: Cyber-capable AI models 27 stories
GPT-6 Cyber built 948 working Linux kernel exploits at $1,279 each, OpenAI says
OpenAI's own research run, in a controlled environment, found and exploited kernel flaws at about $1,279 per exploit. The numbers are OpenAI's; maintainers are being sent the critical findings first.
Check our sources · 6 facts from 2 sourcesOpenAI says a research pipeline it ran in mid-October 2026 produced working local privilege escalation exploits for 948 Linux kernel vulnerabilities, at an API cost of $1,213,430. OpenAI puts that at $1,279 per exploit.
The pipeline analysed 2,308 candidate kernel findings, dynamically reproduced 1,724 of them and made proof-of-concept code for 291 information-leak flaws, according to OpenAI's post of 4 November 2026. It ran in a controlled research environment against a stock desktop operating system, with a human researcher reviewing the results.
OpenAI says it began with the critical vulnerabilities, following guidance from the Linux kernel community, and is working with Patch the Planet partners on reports and patches. Its GPT-6 Cyber announcement of 5 November says it disclosed 304 critical findings to maintainers with candidate patches.
The 4 November post says an upcoming cyber model powered the run. The GPT-6 Cyber announcement ties the 948 exploits to GPT-6 Cyber and says that is more than five times the 166 that GPT-5.6 Sol exploited with the same research harness. All figures are OpenAI's own.
Underlined sentences link to the facts we checked. Dotted words open a short explainer.
Fact 4 of 6 · from OpenAI, 4 Nov 2026
Following guidance from the Linux kernel community, we began with critical vulnerabilities and are working with Patch the Planet partners on further reports and patches.
Your reaction
We count reactions per story and day, never who reacted. The counts help us choose what goes in the monthly issue. If you are signed in, your own page shows yours too.
We count reactions per story and day, never who reacted. They help choose what goes in the monthly issue.
Check our sources
Every sentence above is checked against these 2 sources.
1 A thousand fewer paths to root
Open the source archived copy-
OpenAI's post of 4 November 2026 says its pipeline, run in mid-October 2026 with an upcoming cyber model, dynamically reproduced 1,724 vulnerabilities and generated proofs of concept for 291 information leak vulnerabilities and 948 working local privilege escalation exploits for a commonly used Linux distribution. the company says
citedynamically reproduced 1,724 vulnerabilities and generated PoCs for 291 information leak vulnerabilities and 948 working local privilege escalation (LPE) exploits for a commonly used Linux distribution
-
OpenAI says the run cost $1,213,430 in API credits, or $1,279 per local privilege escalation exploit. the company says
citeThe total run cost $1,213,430 in API credits, or $1,279 per LPE exploit.
-
OpenAI says the prototype identified and analysed 2,308 candidate Linux kernel findings in a controlled research environment, with target virtual machines running a stock desktop operating system. the company says
citeidentified and analyzed 2,308 candidate Linux kernel findings
-
OpenAI says a human security researcher reviews the pipeline results, and that following guidance from the Linux kernel community it began with critical vulnerabilities and is working with Patch the Planet partners on further reports and patches. the company says
citeFollowing guidance from the Linux kernel community, we began with critical vulnerabilities and are working with Patch the Planet partners on further reports and patches.
2 GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents
Open the source archived copy-
OpenAI says it disclosed 304 critical findings to Linux maintainers with candidate patches. the company says
citewe disclosed 304 critical findings to maintainers with candidate patches
-
OpenAI's GPT-6 Cyber post says GPT-6 Cyber reproduced 1,724 vulnerabilities and developed working exploits for 948, more than five times the 166 exploited by GPT-5.6 Sol using the same research harness. the company says
citeGPT‑6 Cyber reproduced 1,724 vulnerabilities and developed working exploits for 948
Every sentence above is checked against these 2 sources. Each fact has its own address you can cite.
Source 1 · OpenAI · 4 Nov 2026
A thousand fewer paths to root
OpenAI's post of 4 November 2026 says its pipeline, run in mid-October 2026 with an upcoming cyber model, dynamically reproduced 1,724 vulnerabilities and generated proofs of concept for 291 information leak vulnerabilities and 948 working local privilege escalation exploits for a commonly used Linux distribution. the company says
dynamically reproduced 1,724 vulnerabilities and generated PoCs for 291 information leak vulnerabilities and 948 working local privilege escalation (LPE) exploits for a commonly used Linux distribution
OpenAI says the run cost $1,213,430 in API credits, or $1,279 per local privilege escalation exploit. the company says
The total run cost $1,213,430 in API credits, or $1,279 per LPE exploit.
OpenAI says the prototype identified and analysed 2,308 candidate Linux kernel findings in a controlled research environment, with target virtual machines running a stock desktop operating system. the company says
identified and analyzed 2,308 candidate Linux kernel findings
Show all 4 factsShow fewer facts
OpenAI says a human security researcher reviews the pipeline results, and that following guidance from the Linux kernel community it began with critical vulnerabilities and is working with Patch the Planet partners on further reports and patches. the company says
Following guidance from the Linux kernel community, we began with critical vulnerabilities and are working with Patch the Planet partners on further reports and patches.
Source 2 · OpenAI · 5 Nov 2026
GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents
OpenAI says it disclosed 304 critical findings to Linux maintainers with candidate patches. the company says
we disclosed 304 critical findings to maintainers with candidate patches
OpenAI's GPT-6 Cyber post says GPT-6 Cyber reproduced 1,724 vulnerabilities and developed working exploits for 948, more than five times the 166 exploited by GPT-5.6 Sol using the same research harness. the company says
GPT‑6 Cyber reproduced 1,724 vulnerabilities and developed working exploits for 948
Topics
The morning email
On the mornings we publish: the three top stories and up to four short ones. Free.
We email you a link to confirm. An issue may include one sponsor, always labelled Sponsored · Advertisement. Our emails count opens and clicks, not who made them. Unsubscribe in one click. What we keep