Org profile: sourced & linked
Organization Profile
Schmidt Sciences
Schmidt Sciences is a philanthropic nonprofit founded by Eric and Wendy Schmidt that funds hypothesis-driven scientific research through targeted grant programs and virtual institutes, with a strong focus on AI safety and advanced computing alongside other scientific domains. Its Science of Trustworthy AI program funds technical research to understand, predict, and control risks from advanced AI systems, in two funding tiers (up to $1M, and $1M–$5M+ over one to three years). Schmidt Sciences proactively solicits proposals through open RFPs and does not accept unsolicited applications.1,2,3
Programs
4 programs on fileMulti-Agent Safety (Scaling AI Safety for a Multi-Agent World)
0 grants tracked
2026 call: closed, closes 2026-08-08
Schmidt Sciences — Science of Trustworthy AI
0 grants tracked
2026 Science of Trustworthy AI RFP: closed, closes 2026-05-17
Schmidt Sciences AI Interpretability RFP
0 grants tracked
AI Interpretability RFP: closed, closes 2026-05-26
Schmidt Sciences RFP for AI-at-Work Field Experiments
0 grants tracked
2026 research agenda: closed, closes 2026-04-10
JanFebMarAprMayJunJulAugSepOctNovDec◆ Today
Multi-Agent Safety (Scaling AI Safety for a Multi-Agent World)2026 call: closed, closes 2026-08-08
0 grants tracked
Schmidt Sciences — Science of Trustworthy AI2026 Science of Trustworthy AI RFP: closed, closes 2026-05-17
0 grants tracked
Schmidt Sciences AI Interpretability RFPAI Interpretability RFP: closed, closes 2026-05-26
0 grants tracked
Schmidt Sciences RFP for AI-at-Work Field Experiments2026 research agenda: closed, closes 2026-04-10
0 grants tracked
Awards
33 on fileShowing 1-25 of 33, newest dated evidence first
RecipientProgramAmountDateEvidence
- Aylin Caliskan — Large Language Model Safety in Inference-Time Motivated ReasoningProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Bo Li — Multiagent-Based T&E Environment ConstructionProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Dan Hendrycks — Utility Engineering and Moral Scaffolding for Safer Reasoning ModelsProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Diyi Yang — Quantifying and Mitigating Privacy Risks in Multi-Agent SystemsProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Ellie Pavlick — Fundamental Limitations of the Test Time Compute ParadigmProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Flammarion — Robust LLM-based Scoring of Agent AlignmentProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Gleave — Deception & Misinformation: Elicitation and TestingProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Hal Daumé — Causally Grounded Inference-Time Intervention for Robust Model AlignmentProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Hashimoto — A Meta-analysis Approach to Understanding LM CapabilitiesProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Jin — Mechanistic Interpretability to Detect Test Set ContaminationProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Kang — Benchmarks for AI Agents and CybersecurityProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Koch — Adaptive Stress Testing for Automated Unsupervised Large Language Model TestingProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Koyejo — Beyond Simple Scaling: A Multi-Dimensional Family of Scaling LawsProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Krueger — What Counts as Contamination? How Generalization Could Confound EvaluationProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Miyazono & Rademaker — Formal Verification of SoftwareProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Narasimhan — Robustness and Controllability of Language Model-Based AgentsProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Owain Evans — Generalization and Hidden Tendencies in LLMsProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Parikh — Will Chain-of-Thought Monitoring Significantly Improve Safety?ProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Raghunathan — Multi-Agent AI Safety via Dynamic GamesProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Sanjeev Arora — Tests of Compositional Generalization as an 'Upper Bound' on AI Safety RisksProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Sap — OpenAgentSafety: Measuring and Mitigating Safety Harms of LLM-based AI Agent InteractionsProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Sergey Levine — Safety in RL-Enabled Goal-Directed AgentsProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Sharon Li — A Conformal Safety Assurance Framework for Large Language ModelsProgramScience of Trustworthy AI (Understanding Safety in AI Systems)Amount–Date–
- Sharon Li — Reasoning with Foresight: Safe Inference via Q-Value Guided DecodingProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
- Shi Feng — Evaluating Preparedness via Model Organism SpectrumProgramScience of Trustworthy AI (Inference-Time Compute)Amount–Date–
4
Programs
33
Awards
33
Grantees
–
Known USD
0
Live rounds
no open or announced rounds on file
Active
Status
Focus Areas
AI safety, AI alignment, interpretability, AI evaluation science, frontier model risk, trustworthy AI, AI oversight, multi-agent risk.
Typical sizeTier 1: up to $1M; Tier 2: $1M–$5M+, each over 1–3 years (2026 Science of Trustworthy AI RFP)
CadenceRFP-driven open calls; does not accept unsolicited proposals