# OpenAI introduces GPT-6 Cyber to find and exploit software vulnerabilities

> Original title: OpenAIs GPT-6 Cyber can both break into systems and propose how to fix them

- Source: The Decoder
- Published: 2026-11-05T20:36:25.000Z
- Bad Influence: https://badinfluence.news/items/eq9wq1kwghleemwqxshufocdg
- Original: https://the-decoder.com/openais-gpt-6-cyber-can-both-break-into-systems-and-propose-how-to-fix-them/

## Summary

OpenAI has introduced GPT-6 Cyber, a model trained to find security vulnerabilities, build working exploits, and suggest patches. OpenAI reports 48.5 percent on ExploitGym, versus 32.1 percent for GPT-5.6 Cyber, and says it stayed within guardrails in 99.93 percent of tests without special safety instructions. The model is initially available through the API and Codex to a select group of Daybreak Red customers, and all benchmark figures come from OpenAI itself.