Physix Frontier · News Briefing Card (TechCrunch · Nov 6, 2026)

OpenAI launches ChatGPT Cyber security model, discloses open-source vulnerability scan results

KEY FACTS

OpenAI released ChatGPT Cyber, a model purpose-built for cybersecurity, available only to the highest tier of users in the Daybreak Red program.
The model scanned the Linux kernel and found 948 potential issues, and discovered 81 novel vulnerabilities in Firefox.
Developers used GPT 5.6 Sol to find a critical Ubuntu server vulnerability within 2 hours, and Debian has released a patch.
OpenAI announced $1 billion to support cybersecurity professionals in critical infrastructure sectors.
OpenAI said only a few thousand security experts participate in the Daybreak program, far fewer than the number of potential attackers.

KEY DATA

948Potential vulnerabilities in the Linux kernel

81Novel vulnerabilities in Firefox

$1.2 millionCost of the Linux vulnerability scan

$1 billionFunding for the cybersecurity support program

PHYSIX OBSERVATION

AI's efficiency at finding vulnerabilities now far exceeds human effort, yet defenders' resources and participation scale are severely insufficient. OpenAI opens up the model while restricting its capabilities at the same time, and this contradiction exposes the imbalance of the security arms race. The patch window for critical infrastructure is narrowing, and the industry needs more open collaboration mechanisms rather than relying solely on charitable scans by a handful of vendors.

Source: TechCrunch report