OpenAI's GPT-6 Cyber finds 81 unknown high-severity Firefox bugs and 588 kernel flaws
Original title: OpenAI’s GPT-6 Cyber ups the ante on vulnerability discovery
OpenAI claims 10 of the model's exploitation techniques were novel; its dashboard shows under 25% of found flaws remediated.
OpenAI introduced GPT-6 Cyber, a restricted-access cybersecurity model it says found 81 previously unknown high-severity vulnerabilities in Firefox and developed working exploits for 948 Linux kernel vulnerabilities, 588 of them previously unknown. Mozilla has begun shipping fixes and Linux findings are under maintainer review. OpenAI says the model stayed within authorized scope in 99.93% of ExploitGym Honeypot cases, and access is limited to vetted Daybreak Red customers. Critics note OpenAI provided no evidence the model speeds patch deployment, and its Patch the Planet dashboard shows a remediation rate under 25%.
Unconfirmed: The vulnerability counts and novelty of the 10 techniques are OpenAI's own claims; the company provided no measurements of faster patch deployment.
- OpenAI says the model produced working exploits for 948 Linux kernel vulnerabilities.
- Researchers flagged 10 exploitation techniques they say external experts had not seen before.
- Mozilla has begun releasing fixes; Linux findings go to maintainers with candidate patches.
- Access is restricted to vetted Daybreak Red customers via API and Codex.
- Vendors
- OpenAI
- Products
- FirefoxLinux kernel
- Organizations
- MozillaCloudflareNoma SecurityDragos
- AI models
- GPT-6 CyberGPT-5.6 CyberGPT-6
From the original article
4 min read at csoonline.com
OpenAI is introducing GPT-6 Cyber, a restricted access model that the company calls its most capable and aligned cybersecurity model yet, “built to help security researchers identify and validate exploitable vulnerabilities, assess their impact, and harden potential patches before they’re released.”
csoonline.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.