OpenAI has introduced GPT-6 Cyber, a new cybersecurity-focused AI model that has already uncovered thousands of issues.
1 editorial report · 1 verified social mention. The most authoritative report leads while later evidence completes the story.
OpenAI says GPT-6 Cyber scored 48.5% on ExploitGym and helped produce 948 Linux privilege-escalation exploits plus 81 high-severity Firefox findings. Mozilla advisories corroborate real fixes, but most performance claims remain company-reported and access requires separate Daybreak Red approval. GPT-6 Cyber Raises Exploit Capability While Keeping Access Restricted
Open mention