Skip to main content
Rase AI Intelligence Newsletter

Safety & governance Model or AI service

OpenAI introduces GPT-6 Cyber for defensive security research

OpenAI announced GPT-6 Cyber as a model for authorized software security research. Its first release was limited to approved organizations. [1]

What the model does

The announcement describes uses in code review, vulnerability triage and patch development. OpenAI says security teams helped test the model in defensive settings. [1]

Why it matters

Software maintainers and security teams can use this class of tool to prioritize reports and review candidate fixes as part of their existing processes. [1]

Key dates

  1. OpenAI announces GPT-6 Cyber [1]

Model limitations

The November announcement describes limited initial access. Security findings and proposed fixes still need human verification. [1]

Sources and verification

Catalog evidence checked How we verify records

  1. Primary source GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents | OpenAI

Related profiles

Safety & governance

Digital Omnibus on AI (AI Act amendments)

European Commission / Parliament / Council

The amendments change elements of AI Act implementation, including distinct future application dates for specified high-risk systems.

High confidenceEvidence checked

Safety & governance

EU General-Purpose AI Code of Practice

European Commission (AI Office)

Provides voluntary guidance on transparency, copyright, and safety and security for providers of general-purpose AI models.

High confidenceEvidence checked