Published in December 2025 by Stanford HAI's Center for Research on Foundation Models, in collaboration with researchers from Berkeley, Princeton, and MIT, this third annual index scores major foundation-model developers on transparency practices. The 2025 edition, for the first time, includes Alibaba, DeepSeek, and xAI among the 13 developers assessed, and finds that industry-wide transparency actually regressed after the prior year's gains.
As one of the only systematic, repeated benchmarks of AI developer transparency specifically, the index has become a widely cited reference point in policy discussions of AI disclosure requirements.