BETA RELEASE

Summary

An expert-led evaluation of eight leading AI companies across six safety and security domains, including risk assessment and existential safety.

Key quotes

Existential safety remains the industry’s core structural weakness
A clear divide persists between the top performers (Anthropic, OpenAI, and Google DeepMind) and the rest of the companies reviewed

The report utilizes a panel of eight AI experts to score companies using a US GPA-style grading system. It evaluates indicators such as dangerous capability evaluations, whistleblower protections, and robustness benchmarks.