GPT‑6 Cyber: Advancing the frontier of aligned cybersecurity agents
What happened
OpenAI introduced GPT‑6 Cyber, a new cybersecurity model designed to find and validate exploitable vulnerabilities while staying within researcher‑defined scope. The model shows a 51% relative improvement on the ExploitGym benchmark, scoring 48.5% versus 32.1% for its predecessor GPT‑5.6 Cyber, and it avoided out‑of‑scope exploits in 99.93% of tests without a prompt and 100% with a prompt. It has already helped uncover thousands of exploits in critical codebases like Linux and Firefox, identifying 81 previously unknown high‑severity Firefox bugs and assisting teams at Cisco, Calif and CodeRabbit. GPT‑6 Cyber is classified as Critical in cybersecurity under OpenAI’s Preparedness Framework, and it powers the Daybreak Red workflow for advanced exploit development and authorized testing.
Why it matters
For everyday users, the model’s ability to securely discover and patch software flaws could lead to more robust and safer applications they rely on. Its strong alignment safeguards reduce the risk of the AI being misused for malicious hacking.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from OpenAI
Introducing ChatGPT WalletOpenAI · 13 Nov 2026Shield helps technicians hand off legacy portal work with OpenAIOpenAI · 13 Nov 2026OpenAI launches ChatGPT Wallet, AI agents pay for purchaseOpenAI · 13 Nov 2026OpenAI disbands its robotics team as its general models start running robotsOpenAI · 13 Nov 2026Yang-Mills, OpenAI proof, and the mass gap problemOpenAI · 13 Nov 2026This page was collected from openai.com and summarised automatically by AIPulsee's pipeline. See how it runs.