OpenAI Launches GPT-6 Cyber, Expanding AI Vulnerability Research and Finds 1000+ Exploits
OpenAI has released GPT-6 Cyber, an AI model specialized for authorized security research that has demonstrated significant improvements in vulnerability discovery. While the model shows high performance in identifying exploits, it raises questions about the balance between automated security testing and the risks of misuse.
Why it matters
The advancement of AI-driven cybersecurity tools represents a double-edged sword for digital infrastructure, necessitating new frameworks for risk management and automated defense.
OpenAI has introduced GPT-6 Cyber, a specialised artificial intelligence model for authorised security research, reporting advances in vulnerability discovery and exploit development alongside tighter controls intended to keep agents within approved boundaries.
Announced on 5 November, the model has already helped researchers develop 948 Linux privilege-escalation exploits and identify 81 previously unknown Firefox vulnerabilities classified as high severity, according to OpenAI’s announcement . Access is initially restricted to selected Daybreak Red customers through Codex and the API.
The release brings a central question for security leaders into sharper focus: how quickly can organisations turn increasingly automated vulnerability research into verified fixes, while containing the risks introduced by agents capable of conducting that research?
OpenAI reports that GPT-6 Cyber achieved 48.5% on ExploitGym, compared with 32.1% for GPT-5.6 Cyber—a gain of 16.4 percentage points, or approximately 51% relative improvement.
Headlinne story · 3 newsrooms
OpenAI Releases New Cybersecurity Model With Strict Safety Controls
What every outlet reported, where they differ, and how each framed it →
Also covering this story
5 other newsrooms covered this event. We read each version separately.
OpenAI launches GPT-6 Cyber for approved security teams
OpenAI launches GPT-6 Cyber with 'Critical' cybersecurity rating
OpenAI launches GPT-6 Cyber with 'Critical' cybersecurity rating
OpenAI says its new cybersecurity model is better at not hacking other companies
OpenAI Launches GPT-6 Cyber After AI Helps Find 81 Firefox Flaws, 948 Linux Exploits
Get smarter about the news
Sign up free for a feed built around what you actually care about, Dive Deeper research on any story, and the full text of every article.
Create free accountAlready have an account? Sign in