Adversarial testing of AI systems to discover failure modes, safety issues, and vulnerabilities, both manual and automated.
Organizations
7
Grants
4
Total Funding
$198K
Risks Addressed
1
Cluster: Evaluation
Parent Area: AI Evaluations
Organizations3
| Organization | Role |
|---|---|
| Anthropic | active |
| Google DeepMind | active |
| OpenAI | active |
Grants4
| Name | Recipient | Amount | Funder | Date |
|---|---|---|---|---|
| UC Berkeley — AI Red-teaming Bootcamp | University of California, Berkeley | $100K | Coefficient Giving | 2025-01 |
| Meta level adversarial evaluation of debate (scalable oversight technique) on simple math problems (MATS 5.0 project) | Yoav Tzfati | $62K | Long-Term Future Fund (LTFF) | 2024-01 |
| SoGive does EA analysis. We red-team EA analysis and conduct EA research to support donors. We also support EA talent | SoGive | $18K | Centre for Effective Altruism | 2022-07 |
| SoGive does EA analysis. We red-team EA analysis and conduct EA research to support donors. We also support EA talent | SoGive | $18K | Centre for Effective Altruism | 2022-07 |
Funding by Funder
| Funder | Grants | Total Amount |
|---|---|---|
| Coefficient Giving | 1 | $100K |
| Long-Term Future Fund (LTFF) | 1 | $62K |
| Centre for Effective Altruism | 1 | $18K |
| Centre for Effective Altruism | 1 | $18K |
Tags
red-teamingadversarialsafety-testing