Overview
Perrin Giving funds think tanks and technical projects that put AI to work for people.
Perrin Giving funds think tanks and technical projects that put AI to work for people, including work we are not positioned to do ourselves.
We fund people close to the problem. We do not screen on stage or size: a small team that has already started is as eligible as an established institution.
Every grant is published with the reasoning behind it.
What we fund
- 01
Think tanks
Independent research on the positive uses of AI and on the legislation forming around it. Unrestricted wherever the recipient can accept it, so that funding never shapes the findings.
- 02
Technical projects
Working systems rather than proposals about them: evaluation and testing tools, appeals and disclosure systems, and the practical components that are hard to fund.
- 03
The work between them
Standards drafting, replication, and the legal and procurement work that decides whether a working technology is ever allowed to be used. This work is rarely funded, and it often decides whether a working technology can be used at all.
Grant terms
- Recipients
- Think tanks, technical projects, and individuals who have already started
- Instrument
- Grant, unrestricted wherever the recipient can take it
- Decision
- Two weeks from first conversation
- Reporting
- One page, twice a year
- Published
- Every grant, with the reasoning behind it
The evidence
Access, and what gets funded.
- World
- High income
- Low & middle income
Source: World Bank, Global Findex Database 2025, Findex 2025, API last updated 2025-10-06. CC BY 4.0. Survey years only.
View the data
| World | High income | Low & middle income | |
|---|---|---|---|
| 2011 | 50.6 | 87.5 | 41.6 |
| 2014 | 61.7 | 92.2 | 54.7 |
| 2017 | 69.3 | 93 | 64 |
| 2021 | 73.8 | 95.8 | 69 |
| 2024 | 78.7 | 94.9 | 75.4 |
Source: Federal Deposit Insurance Corporation (FDIC), National Survey of Unbanked and Underbanked Households, 2023, November 2024. Public domain.
View the data
| % of households | |
|---|---|
| 2009 | 7.6 |
| 2011 | 8.2 |
| 2013 | 7.7 |
| 2015 | 7 |
| 2017 | 6.5 |
| 2019 | 5.4 |
| 2021 | 4.5 |
| 2023 | 4.2 |
- Grants
- Contracts
- Other Transaction Agreements
Source: Stanford HAI, AI Index 2026, fig. 8.5.2, April 2026 (9th edition). Redrawn from the AI Index public data, CC BY-ND 4.0.
View the data
| Grants | Contracts | Other Transaction Agreements | |
|---|---|---|---|
| 2016 | 0.079 | 0.056 | 0.001 |
| 2017 | 0.142 | 0.134 | 0 |
| 2018 | 0.269 | 0.209 | 0.012 |
| 2019 | 0.43 | 0.279 | 0.101 |
| 2020 | 0.699 | 0.461 | 0.078 |
| 2021 | 1.665 | 0.703 | 0.072 |
| 2022 | 3.349 | 0.581 | 0.069 |
| 2023 | 4.026 | 0.643 | 0.197 |
| 2024 | 5.047 | 0.686 | 0.124 |
- Security & safety
- Fairness & bias
- Transparency & explainability
- Privacy & data governance
Source: Stanford HAI, AI Index 2026, fig. 3.4.2, April 2026 (9th edition). Redrawn from the AI Index public data, CC BY-ND 4.0. A paper can carry several topics.
View the data
| Security & safety | Fairness & bias | Transparency & explainability | Privacy & data governance | |
|---|---|---|---|---|
| 2019 | 162 | 57 | 89 | 39 |
| 2020 | 168 | 98 | 134 | 124 |
| 2021 | 215 | 150 | 189 | 150 |
| 2022 | 285 | 169 | 231 | 85 |
| 2023 | 276 | 212 | 393 | 213 |
| 2024 | 521 | 408 | 355 | 186 |
| 2025 | 641 | 462 | 405 | 248 |