FMF Launches a Framework for Rating AI Model Jailbreaks
The Frontier Model Forum released version 1.0 of its Jailbreak Rating Framework, a proposal modeled loosely on CVSS that rates LLM jailbreaks by capability gain, capability scope, deployment ease, and discovery ease. Tom’s Hardware reports that the framework combines severity and likelihood into a rating from low to critical, with the capability-gain score constraining the other dimensions. FMF says the cross-vendor scale could help teams developing patches, but version 1.0 remains under review, and the group advises against relying on it while comments and suggestions are collected through December 21.
FMF intends the framework to provide a common cross-vendor method for comparing LLM jailbreaks and their potential to enable cyber operations beyond existing tools.
In recent months, high-profile headlines about AI models letting loose from their safeguards dominated headlines. Now, the Frontier Model Forum (FMF) has released a new Jailbreak Rating Framework that aims to track how easy it might be for a given AI model to get around safeguards. The framework also estimates the scale of the collateral damage, should an AI model wreak cyber-havoc across the internet.
The project started in 2023 with the support of four founding companies: Anthropic, Google, Microsoft, and OpenAI. The published Jailbreak Rating Framework is a rough analog to the well-known CVSS security rating system, but specifically targeted at classifying the type and severity of LLM jailbreaks across various axes and multiple frontier LLM providers.
Many AI companies and evaluation entities already have their own rating systems, but the FMF intends for its scale to be a cross-vendor standard that teams developing patches can rely on.
The FMF's approach is to have a classification method for jailbreaks that can make a model perform cyber-warfare operations beyond what's already out, across both the regular and black market. Version 1.0 of the proposal is a work-in-progress, and the group is taking comments and suggestions on it until December 21, noting that relying on its system in the meantime is inadvisable.
#Frontier Model Forum#Jailbreak Rating Framework#Anthropic#Google#Microsoft#OpenAI