SaferAI Ratings 2026

SaferAI scores twelve developers’ published frameworks on 65 weighted criteria across four dimensions: risk identification, risk analysis and evaluation, risk treatment, and risk governance. July 2026 scores ran from 35% (Anthropic) to 8% (Cohere). The methodology is published in Evaluating AI Providers’ Frontier Safety Frameworks and there is a companion report on emerging best practices (July 2026).