OpenAI backs Frontier Model Forum jailbreak rating framework for cyber threats
OpenAI said on 6 November 2026 that it partnered with the Frontier Model Forum on a shared framework to rate when bypassing AI safeguards could enable serious cyber harm and to prioritize stronger protections.
OpenAI announced on X on 6 November 2026 that it worked with the Frontier Model Forum and member firms on a framework for rating cyber jailbreaks.
“We partnered with @fmf_org and its member firms on a new framework to help identify when bypassing an AI system’s safeguards could enable serious cyber harm, and prioritize stronger protections.”
— OpenAI (@OpenAI), 6 November 2026
The longer note quotes Sasha Baker, OpenAI’s head of national security policy, calling the work a step toward a common methodology for comparing the severity of cyber jailbreaks. The Frontier Model Forum update describes Version 1.0 of the Jailbreak Rating Framework for Frontier Cyber Threats.
Background
Earlier stories on this thread.
What happened next
Later stories on this thread.
Shared rating, not a new model
This is a standards and policy story, not a product launch. OpenAI frames the framework as a way for AI and security teams to build a clearer evidence base and prioritize safeguard work. Severity labels and scoring dimensions come from the Forum’s published specification; they are industry self-definitions, not a regulator mandate.
The public anchors remain the OpenAI X post and the FMF announcement.
This report is not legal, financial, or health advice. See the disclaimer.
Primary sources
- OpenAI on X: FMF jailbreak framework OpenAI on X
- Introducing the FMF Jailbreak Rating Framework for Frontier Cyber Threats Frontier Model Forum