OpenAI released ChatGPT Cyber, a cybersecurity model restricted to vetted researchers in its Daybreak Red program.
The model flagged 81 novel Firefox vulnerabilities and 948 potential issues in the Linux kernel. It also helped uncover a severe Ubuntu/Debian server flaw tracked as CVE-2026-109518.
OpenAI is committing $1B to support Daybreak participants working on critical infrastructure.
https://techcrunch.com/2026/11/05/openai-released-a-new-cybersecurity-model-is-that-enough/
