OpenAI Launches ChatGPT Cyber After AI Uncovers 948 Linux Kernel Flaws

3 articles · Updated · TechCrunch · Nov 5

ChatGPT Cyber is being released to top-tier Daybreak Red users after OpenAI highlighted AI-assisted bug hunting that surfaced a severe Ubuntu-related server takeover flaw and hundreds of other issues.
Max Justicz said GPT-5.6 Sol found the Ubuntu vulnerability in about two hours after he prompted it to search code linked to a bug he had previously reported; Debian has already patched it and no exploitation is known.
OpenAI said the same cybersecurity model helped scan and prepare patches for 948 potential Linux kernel issues and identified 81 novel Firefox vulnerabilities, underscoring how quickly frontier models can expand defenders’ workload.
Daybreak still reaches only thousands of security professionals, while Justicz said he spent about $5,000 on tokens and subscriptions and criticized recent limits on Daybreak Blue that may blunt defenders more than attackers.
OpenAI has paired the launch with a $1 billion cybersecurity initiative, but its disclosed $1.2 million Linux bug-hunt spend suggests the race to patch critical infrastructure may demand greater urgency.