Skip to content
OpenAI's GPT-6 Cyber Model Rated Critical for Cybersecurity

OpenAI's GPT-6 Cyber Model Rated Critical for Cybersecurity

First seen 9 Nov 2026, 12:37 UTC • •
API

Article Content

Browse articles
ThreatCluster AI
ThreatCluster •November 9, 2026 at 12:38 UTC
  • •GPT-6 Cyber rated Critical, indicating high risk in cybersecurity.
  • •Model found 81 high-severity flaws in Firefox and 948 exploitable vulnerabilities in Linux.
  • •Access is restricted to vetted security teams for offensive security tasks.

OpenAI introduced GPT-6 Cyber on November 5, 2026, a model designed to identify and exploit software vulnerabilities. It has been rated as Critical by OpenAI's Preparedness Framework, indicating the highest risk level. The model has already found 81 high-severity vulnerabilities in Firefox and reproduced 1,724 vulnerabilities in the Linux kernel, with 948 exploitable. OpenAI has reported 304 critical findings to Linux kernel maintainers, and patches are in development. Access to GPT-6 Cyber is limited to vetted security teams under the Daybreak Red program. The model's predecessor, GPT-5.6 Cyber, was rated High and had a significantly lower completion rate for exploit tasks. OpenAI's internal tests show that GPT-6 Cyber successfully completed 48.5% of tasks on ExploitGym, compared to 32.1% for GPT-5.6 Cyber. Concerns have been raised about the implications of such powerful tools in the hands of malicious actors.

Start a free Starter trial for enhanced analysis

Ask AI about this cluster

Updated 1d ago How this analysis works

Timeline

2026-10-27
Firefox 159 security advisory published
Mozilla published an advisory crediting OpenAI for 41 of 167 fixed vulnerabilities in Firefox.
Notebookcheck
2026-11-05
GPT-6 Cyber introduced
OpenAI launched GPT-6 Cyber, a model for identifying and exploiting software vulnerabilities.
Notebookcheck
Recent
Critical findings reported to Linux maintainers
OpenAI reported 304 critical vulnerabilities to Linux kernel maintainers, with ongoing patch development.
Notebookcheck

More articles in this cluster (2)

Following this threat?

Track Fedora in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.

Free account · no card needed

Extracted Entities (5)

1 / 2

Common questions

What vulnerabilities did GPT-6 Cyber find?
GPT-6 Cyber identified 81 high-severity vulnerabilities in Firefox and 948 exploitable vulnerabilities in the Linux kernel.
Who has access to GPT-6 Cyber?
Access is limited to vetted security teams participating in OpenAI's Daybreak Red program.
What is the current status of the vulnerabilities?
Patches for the identified vulnerabilities are in development, but exploitation has not been confirmed.