organization · Sergey Levine, UC Berkeley
Technical AI safety research
| Coefficient Giving | $1,450,016 | 63% | 1 |
| FTX Future Fund | $600,000 | 26% | 1 |
| Schmidt Sciences | $250,000 | 11% | 1 |
| Date ↓ | Funder | Via | Amount | Cause | Source | Purpose |
|---|---|---|---|---|---|---|
| 2025 | Schmidt Sciences | ~$250,000* | Technical AI safety research | schmidt_sciences | Project: Safety in RL-Enabled Goal-Directed Agents. Inference-time compute safety cohort; per-project amount undisclosed (RFP cap $500K). — Science of Trustworthy AI — Inference-Time Compute (2025 cohort) | |
| 2022 | FTX Future Fund | $600,000 | Technical AI safety research | ftx_future_fund | We recommended a grant to support a project to study how large language models integrated with offline reinforcement learning pose a risk of machine deception and persuasion. — Artificial Intelligence | |
| Oct 2017 | Coefficient Giving | University of California, Berkeley | $1,450,016 | Technical AI safety research | coefficient_giving | AI Safety Research (Sergey Levine, Anca Dragan) |
* Estimated from Schmidt Sciences' reporting of total given out in the funding round; grant-level amounts were not provided.