Launch3 sources6 comments in the community

OpenAI’s new GPT-6 Cyber is better at hacking and better at staying in its lane

The New Stack · By Frederic Lardinois · 17h ago

OpenAI on Thursday launched GPT-6 Cyber , the latest release in its series of cybersecurity-focused models. The company ranks GPT-6 Cyber’s ability to identify vulnerabilities and develop exploits as ‘critical,’ up from ‘high’ for its predecessor. Just like with previous Cyber…

Read the full story at thenewstack.io ↗

The Brief

OpenAI launched GPT-6 Cyber on 2026-11-05, the latest model in its cybersecurity-focused series.
OpenAI rates GPT-6 Cyber's ability to identify vulnerabilities and develop exploits as 'critical', up from 'high' for its predecessor.
The model is designed to find exploitable vulnerabilities and develop fixes, and is trained to stay within authorized scope.

Why it matters

Security teams doing authorized vulnerability work get a model OpenAI rates as more capable at finding and exploiting vulnerabilities, with scope restriction as its stated safeguard.

Who’s affected

Security teams doing authorized vulnerability work
Developers using OpenAI models for security testing

#openai
#ai-security
#vulnerability-research
#cybersecurity
#llm

What developers are saying

Based on 6 comments · Mixed reactions

The two Hacker News commenters question the timing of releases and access restrictions. One suggests OpenAI and Anthropic delay new models until the other releases theirs. The other asks why availability of 'Daybreak Blue' is delayed when high-impact CVEs appear several times a week, and suggests vulnerability finding could be delegated to another model.

Competitive release timing50%
Delayed access to security tooling50%
Alternative open models25%

In favour

Concerns

− Delaying availability is pointless when high-impact CVEs are being disclosed frequently
− Model releases appear timed around rival labs rather than readiness