Schmidt Sciences — Science of Trustworthy AI

Schmidt Sciences' Science of Trustworthy AI program funds technical research to understand, predict, and control risks from advanced AI systems, across three aims: characterizing/forecasting misalignment in frontier models, developing generalizable measurement and intervention, and overseeing superhuman-capability and multi-agent systems. Awardees also receive compute, software engineering support, and frontier-model API credits. Funded in two tiers (up to $1M; $1M-$5M+ over 1-3 years) across sub-tracks including "Understanding Safety in AI Systems" (23 awards) and "Inference-Time Compute" (10 awards).