Testing AI systems specifically for dangerous capabilities like CBRN knowledge, cyber offense, autonomous replication, and persuasion.
Organizations
4
Grants
8
Total Funding
$18M
Cluster: Evaluation
Parent Area: AI Evaluations
Grants8
| Name | Recipient | Amount | Funder | Date |
|---|---|---|---|---|
| RAND — AI Evaluation and Testing | RAND | $10M | Coefficient Giving | 2025-09 |
| UC Berkeley — Cyberoffense Benchmark | University of California, Berkeley | $3.4M | Coefficient Giving | 2025-06 |
| Apollo Research — General Support | Apollo Research | $2.2M | Coefficient Giving | 2024-05 |
| Owain Evans Research Group — AI Evaluations Research | Effective Ventures Foundation USA | $1.2M | Coefficient Giving | 2023-05 |
| Modulo Research — AI Safety Research | Modulo Research | $408K | Coefficient Giving | 2023-08 |
| Forecasting Research Institute — Tripwire Capability Evaluations | Forecasting Research Institute (FRI) | $359K | Coefficient Giving | 2024-10 |
| Isolating CBRN Knowledge in LLMs for Safety - Phase 2 (Research) | Krishna Patel | $150K | Manifund | 2025-12-17 |
| Surge AI — Data Production for AI Safety Research | Surge AI | $126K | Coefficient Giving | 2023-09 |
Funding by Funder
| Funder | Grants | Total Amount |
|---|---|---|
| Coefficient Giving | 7 | $18M |
| Manifund | 1 | $150K |
Tags
evaluationsdangerous-capabilitiescbrn