AI EQUITY PROJECT 2026
AI is widely used. Mission impact is still unclear.
By 2026, AI use had become widespread—formally and informally. Governance structures were more visible than in earlier years, but adoption was still moving faster than some of the equity, accountability, and community-participation structures surrounding it.
880
Organizations
~70%
Report AI use
12%
Org-wide guidance
57%
AI impact unclear
CONTEXT
Three years into the project
In 2024, the sector was trying to understand what AI meant. In 2025, organizations moved into experimentation and began designing the structures around it. In 2026, the questions became harder because AI was no longer hypothetical.
The 2026 survey gathered responses from 880 organizations, primarily across the U.S. and Canada, between May 1 and July 30. We asked about AI use, readiness, organizational culture, governance, decision-making, funding, vendor selection, accountability for harm, community participation, and philanthropy’s role in shaping adoption.
Across three years, the story is a movement from fog, to naming, to adoption—with AI adoption moving faster than some of the equity structures surrounding it. The next phase is not simply about using more AI. It is about deciding when to use it, who should shape those decisions, and what accountability looks like when the consequences become real.
DEFINITION
What do we mean by AI Equity?
AI Equity is the development, funding, deployment, and use of AI in ways that distribute power, benefits, risks, and accountability more fairly—and center the voice, agency, and protection of people and communities, including those historically underserved.
In practice, that means transparent, accountable, and participatory AI decisions; reducing inequitable harm; distributing benefits and decision-making power more fairly; and making responsibility clear when something goes wrong.
FINDINGS
What did we learn?
01
AI adoption continues to outpace governance
Nearly 70% of respondents reported some organizational AI use. But only 12% said that use was guided by organization-wide practices or expectations. Just 21% had a formal process establishing who would be accountable if AI caused harm; another 20% said accountability remained unclear, and 21% were still designing a process.
AI is often entering organizations through individual initiative rather than deliberate institutional decisions. Staff experiment, tools spread, and only afterward do organizations begin discussing procurement, policy, accountability, or harm. AI can become organizational practice before the organization has decided what responsible practice requires.
02
Organizations feel more ready—but not collectively aligned
Forty-eight percent rated their organizations as moderately, very, or fully ready to integrate AI. But readiness did not mean shared organizational alignment. Only 37% described the tone around AI as enthusiastic or cautiously optimistic.
Others described organizations divided between enthusiasm and resistance, unclear because leadership had not communicated a position, compliant but reluctant, or actively avoidant. Some teams have approved platforms and coordinated experimentation; others have individual staff using consumer tools while leadership remains uncertain or silent. Readiness is growing faster than collective alignment.
03
AI is inexpensive to start—and expensive to understand
Most nonprofits are not beginning their AI journeys through dedicated grants. They are absorbing AI through free tools, existing subscriptions, internal software budgets, open-source tools, and even staff members’ own money.
That makes experimentation relatively easy to begin. Understanding whether those experiments create meaningful mission value is much harder. Difficulty measuring AI impact rose from 35% in 2025 to 57% in 2026, becoming the most commonly selected challenge to securing AI-related funding.
The challenge is not just technical evaluation. Organizations are still learning how to distinguish increased output from improved outcomes—and how to define responsible AI success in the first place.
04
Philanthropy is shaping AI without consistently stewarding it
Forty-three percent of respondents said AI had not come up in communication with their funders or donors during the previous year. But where funders were involved, that involvement did not always translate into meaningful support.
Twenty-one percent said funders encouraged or required AI use, 19% said funders used AI to review or process grant applications, and 12% had been required to report on AI use. Yet only 17% had been offered AI-specific funding and 14% had been offered responsible-AI learning opportunities.
Encouraging adoption is not the same as paying for governance, evaluation, staff time, or community participation. Philanthropy is already shaping the environment around AI; the question is whether it will also help steward the infrastructure needed to navigate that environment responsibly.
FUNDING
How nonprofits are paying for AI
HOW AI IS BEING ACCESSED OR FUNDED | 2026 |
|---|---|
Free or consumer accounts | 36% |
Existing software subscriptions | 29% |
Organizational software budgets | 28% |
Staff paying personally without reimbursement | 22% |
Open-source tools | 18% |
AI-specific grant funding | 10% |
The pattern matters: AI can be cheap enough to enter an organization without a formal funding decision, while the work required to evaluate, govern, and sustain it remains underfunded.
If funding were available, what would nonprofits want it to cover?
PRIORITY | 2026 |
|---|---|
Staff training and AI literacy/fluency | 49% |
AI governance policy development | 31% |
Hire an AI / Data Equity consultant or staff role | 28% |
Adoption and change-management strategy | 27% |
Community engagement processes around AI | 23% |
Pilot an AI tool with proper evaluation | 22% |
Research into AI harms specific to our community | 20% |
General operating support for AI readiness | 20% |
The request is not simply “buy us AI.” Nonprofits are asking for the capacity to understand it, govern it, evaluate it, and involve people in the decisions around it.
THREE YEAR SAMPLES
What three years of data tell us
Across 2024–2026, adoption has become more normal, policies have become more visible, and the sector has developed a stronger vocabulary for risk and responsibility. But the longitudinal picture is not a simple story of progress.
CUSTOM INDICATOR | 2024 | 2025 | 2026 |
|---|---|---|---|
AI Equity Awareness Index | 49% | 67% | 68% |
Equity Practice Ratio | 92% | 62% | 69% |
Policy Preparedness Ratio | 10% | 20% | 29% |
AI Caution Ratio | 83% | 80% | 84% |
The 2026 Readiness Conversion Ratio is intentionally not shown because the underlying AI-use and readiness questions changed enough that direct comparison with 2024–2025 would be unreliable. The project keeps longitudinal measures only where the questions remain sufficiently comparable.
OUTPUTS
What did we produce from this work?
Three years of research pointed to a need for more than findings. The sector needed a shared structure for asking better questions about AI—before, during, and after adoption. In 2026, that learning became the AI Equity Framework.
AI Equity Workbooks
Reflection & discussion tools for nonprofits, funders, and technology partners to examine power, participation, resources, accountability, and impact.
AI Equity Checklist
A practical starting point for asking critical, ethical, and strategic questions before adopting, expanding, funding, or recommending an AI tool.
AI Equity Framework
A shared structure connecting Data Equity, Accountability, Transparency, Resource Equity, Co-Design and Participation, and Power and Self-Determination.
OUTPUTS
What does the AI Equity Framework look like?
Three years of research pointed to a need for more than findings. The sector needed a shared structure for asking better questions about AI—before, during, and after adoption. In 2026, that learning became the AI Equity Framework.
01
Data Equity
Whose data are represented, protected, excluded, or put at risk—and whether data practices are fair enough to support responsible AI.
04
Resource Equity
Who has the money, time, staff capacity, training, and technical support to participate safely in AI adoption.
02
Accountability
Who owns the decision, who responds when something goes wrong, and whether people have a meaningful path to raise concerns or appeal.
04
Co-Design & Participation
Whether staff, communities, and people affected by AI decisions have meaningful opportunities to shape them.
03
Transparency
What data, assumptions, limitations, intended uses, risks, impacts, and decision criteria are visible to the people affected by the system.
04
Power & Self-Determination
Whether people retain agency, refusal, human review, and influence over how AI changes decisions that affect them.
IMPLICATIONS
What the findings ask of the sector
For funders
Fund the infrastructure around AI—not only licenses and pilots. Support training, governance, leadership education, community engagement, and evaluation. Apply the same rigor to AI vendors that you expect from grantees, and disclose when AI influences screening, scoring, or funding decisions.
For nonprofit leaders and teams
Conduct readiness and equity reviews before scaling AI. Decide who owns governance, who approves tools, who investigates concerns, and who is accountable when harm occurs. Train staff in AI and Data Equity fundamentals and make equity questions part of routine technology decisions.
For sector networks, associations, and policymakers
Create practical policies, procurement questions, evaluation tools, peer-learning spaces, incident-sharing mechanisms, and shared expectations for disclosure, consent, human review, privacy, worker protection, and community participation.
For technology developers
Design for the realities of nonprofit work: small teams, limited budgets, uneven technical capacity, and sensitive community data. Make limitations visible, build in refusal and human review, create clear ways to report harm, and show how user and community feedback changed the product.
A DELIBERATE CHOICE
Why we still do not create an AI Equity Score
Three years into the project, we still choose not to reduce AI Equity to a single score. One number would hide the tensions the data asks us to notice.
​
An organization can feel ready while lacking shared guidance. It can have a policy without clear accountability. It can use AI widely while communities have little voice in how it is adopted. A high score could reward visible activity without revealing who holds power, who carries risk, or whether caution is being respected.
​
AI Equity is not a race toward a perfect score. It is an evolving practice of examining readiness, responsibility, power, participation, and impact with greater depth and honesty.
CONTINUE THE WORK
Research should help the sector act
The AI Equity Project has now created three years of sector learning. The next chapter may include deeper qualitative research, new tools, wider geographic participation, more community-led learning, and stronger dialogue among nonprofits, funders, technologists, networks, and the communities most affected by AI decisions. The commitment remains the same: research should not simply document what is happening. It should help the sector question, adapt, and act.