Center for AI Safety

San Francisco, California

established25 peopleFounded 2022

Endorsed by:

The Center for AI Safety (CAIS) is a San Francisco-based nonprofit founded in 2022 by Dan Hendrycks and Oliver Zhang. CAIS operates across three pillars: conducting technical and conceptual AI safety research, building the AI safety research field through compute infrastructure, fellowships, courses, and competitions, and advocating for AI safety standards with policymakers and the public. The organization provides free access to a GPU compute cluster supporting hundreds of researchers, publishes foundational benchmarks and safety methods at top ML conferences, and runs educational programs including a widely-used AI safety textbook and course.

Funding Details

Annual Budget: $7,163,607
Monthly Burn Rate: $596,967
Current Runway: -
Funding Goal: -
Funding Raised to Date: $32,984,336
Fiscal Sponsor: -

Theory of Change

CAIS believes that reducing societal-scale AI risk requires a multi-pronged approach. Through technical research, they develop benchmarks, safeguards, and safety methods that make AI systems more reliable and resistant to misuse, providing the scientific foundations the community needs to address safety challenges. Through field-building, they lower barriers to entry for AI safety research by providing free compute infrastructure, running educational programs, and organizing competitions, thereby expanding the number of researchers working on AI safety. Through advocacy, they raise public awareness, advise policymakers, and promote safety standards, ensuring that governance keeps pace with AI capabilities. The causal chain runs from producing concrete safety tools and knowledge, to growing the talent pipeline, to shaping policy and norms, all aimed at ensuring AI development proceeds in ways that reduce catastrophic and existential risk.