RiskRubric Overview
Released: 10/22/2025
Updated: 08/14/2025
RiskRubric is a collaborative initiative between the Cloud Security Alliance (CSA), Harmonic, Haize Labs, and Noma that provides automated, expert-verified safety assessments of publicly available AI models. This dynamic leaderboard helps AI buyers, builders, and regulators evaluate and compare model risk using a standardized, transparent approach.
Each model is scored daily across six key dimensions: Transparency, Reliability, Security, Privacy, Safety, and Reputation. The result is a continuously updated report card that supports safer AI adoption across the ecosystem.
Read this document to learn about:
- How RiskRubric works, including its automated model collection and scoring methodology
- The six categories used to evaluate risk, and why they matter
- The benefits of using RiskRubric to inform AI model decisions
- Why this partnership between CSA, Harmonic, Haize Labs, and Noma makes RiskRubric a trusted industry standard
- How to access the leaderboard and interpret results
Resource unavailable
