Schmidt Sciences 2026 Science of Trustworthy AI Research Grants (Discontinued)

Schmidt Sciences invites proposals for the 2026 Science of Trustworthy AI RFP, funding technical research that advances the science of building trustworthy AI systems. The program addresses three interconnected research aims: understanding why frontier AI systems develop misaligned goals that fail under distribution shift or pressure (Aim 1), creating valid evaluations and interventions to control what AI systems learn (Aim 2), and developing oversight mechanisms for superhuman AI capabilities and managing multi-agent risks (Aim 3). Beyond direct funding, awardees receive computing resources including GPUs and CPUs, software engineering support, API credits with frontier model providers, and access to a research community. The program is open globally and encourages cross-institutional and cross-geographic collaborations. Indirect costs are capped at 10% of total direct costs.

Schmidt Sciences' Science of Trustworthy AI program funds technical research to make frontier AI systems safe and reliable. Grants are offered in two tiers: Tier 1 up to $1M and Tier 2 from $1M to $5M+, each over one to three years, plus in-kind computing resources, software engineering assistance, and API credits. Research addresses three interconnected aims: investigating misaligned objectives in frontier AI under distribution shift, creating valid evaluations and interventions for AI learning control, and developing oversight mechanisms for superhuman capabilities and multi-agent risks. The 2026 cycle closed May 17, 2026 with notifications expected in summer 2026.

Schmidt Futures 2026 Science of Trustworthy AI RFP is a grant from Schmidt Futures that funds technical research improving our ability to understand, predict, and control risks from frontier AI systems while enabling their trustworthy deployment. The program supports work aligned with its published Research Agenda defining scientific scope and priorities. Funding is available in two tiers: Tier 1 up to $1M and Tier 2 from $1M to $5M or more, each for one to three years. The application deadline is May 17, 2026. Eligible applicants include researchers and institutions working on AI trustworthiness and safety.

The Science of Trustworthy AI program from Schmidt Sciences funds technical research to understand, predict, and control risks from frontier AI systems. The 2026 RFP offers two funding tiers: Tier 1 up to $1 million and Tier 2 from $1 million to $5 million or more, each spanning one to three years. The research agenda focuses on three aims: characterizing misalignment in frontier AI, developing generalizable measurements and interventions, and overseeing AI systems with superhuman capabilities including multi-agent risks. Proposals are open globally to universities, national laboratories, institutes, and nonprofit research organizations, with preference for multi-PI collaborations. Schmidt Sciences also provides compute resources, software engineering support, and API credits with frontier model providers. Indirect costs must not exceed 10 percent.