GoKawiilFrontier Model Forum publishes v1.0 Jailbreak Rating Framework for AI models
The Frontier Model Forum, backed by Anthropic, Google, Microsoft and OpenAI, has released a draft scoring system for rating how severe and exploitable AI jailbreaks are. The framework scores jailbreaks on capability gain, scope and ease of deployment, combining them into a single low-to-critical rating, and is open for public comment until December 21.
GoKawiil's interpretation of the reporting above, not reported fact.
A shared standard could let AI developers and security teams compare jailbreak severity across different companies' models rather than relying on inconsistent in-house systems, similar to how CVSS standardized software vulnerability scoring. Because the framework is still a draft under review, its creators caution against using it for real decisions yet, which suggests the industry is still working out how to quantify AI misuse risk consistently.
- FMF, founded by Anthropic, Google, Microsoft and OpenAI, released its Jailbreak Rating Framework v1.0.
- The system scores jailbreaks on severity (capability gain, scope) and likelihood (ease of discovery/use).
- Public feedback is open until December 21, and the framework isn't yet recommended for production use.
Source: tomshardware.com — Bruno Ferreira, 2026-11-10
Published there as: “New framework grades AI jailbreaks across capability gains, scope, and how easy they are to deploy”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.