2026 State of Nonprofit AI Report: Nonprofits Are Using AI Faster Than They're Training for It

AI has quietly become part of the daily routine at 98% of nonprofits, according to a new survey from NTEN and The Bridgespan Group. The catch? Only 32% of staff report having any formal training over the past year.

In short, while teams are already experimenting with AI, the infrastructure to support it is still trying to catch up. If your team is using AI and you're still working out the guardrails, this report gives you a benchmark and a place to start.

Details

What NTEN and Bridgespan released

The 2026 State of Nonprofit AI Adoption and Governance Report is by David Figueroa and Jacob Dwelle. NTEN is a community of nonprofit professionals that provides education and research on nonprofit technology. The Bridgespan Group is a global nonprofit that has advised social change leaders since 2000.

• Who responded: 917 nonprofit staff and executives, mostly in the U.S., with many from around the world
• When: the survey ran from late April to mid-June 2026, and the report came out September 10, 2026
• Cost: free with an NTEN account
• Worth knowing: the survey went out through NTEN and Bridgespan networks, so respondents may be more tech-engaged than the sector overall

What they found

• Shadow use is common, including among leaders. 53% of respondents used AI informally, on personal devices, or outside organizational guidance. Executives did it more often than staff, 57% versus 49%.
• The guardrails are still being built. 57% of executives say their organization has no AI budget. Only 22% say a plan for managing AI risk is fully in place, and the same share say that about staff training on safe use.
• Few harms reported, for now. Only 6% of respondents know of an AI-related incident that caused harm at their organization in the past year. The authors caution that this counts harm someone noticed, not actual risk. In its September 16 analysis, Nonprofit Quarterly asks the sharper question: is the sector managing AI well, or do most organizations lack a way to know when something goes wrong?
• Staff and executives feel differently about it. 63% of executives feel positive about AI use at work, compared with 39% of staff.
• Both groups agree on the fix. 62% of staff say clear organizational policies would help them significantly. Executives name staff training (47%) and clear guidance (45%) as top ways to expand responsible use.

Why this matters for your nonprofit

• Executive directors and boards get a benchmark for deciding where to invest.
• Staff asking for clearer guidance now have data behind that request.
• Funders can see where support is thin. Among organizations with restricted AI grants, staff training (24%) and governance guardrails (10%) were the least-funded uses.

How to get started

What you'll need

• AI Playbook’s summary of the report to quickly review key highlights
• A free NTEN account, to sign in and get the full report
• Reading time. The report runs 81 pages, and many findings are split by role and by budget size (under $10 million and $10 million or more).
• One person to own the follow-up. On a small team, that's usually the executive director or whoever handles operations or IT.

Steps

1. Get the report. Sign in on the report page with your NTEN account. The page also offers graphics of the key charts, ready to share.
2. Start with the three trends. Page 5 sums up the findings: interest in AI has outpaced the infrastructure to support it, staff and executives experience adoption very differently, and the sector already agrees on the fix. Then read the Shadow use and Training and support sections.
3. Check your own guardrails. The report tracks six governance pieces: written guidance on safe use, rules on what data can go into AI tools, a named owner for oversight, a process to approve tools and vendors, a plan for managing AI risk, and staff training on safe use. Mark each one the way the survey did: in place now, in development, or not in place.
4. Ask your staff, not just your leadership team. Staff and executives in this survey saw the same rollout very differently. Ask your staff what would help them use AI with confidence, then compare their answers with the report's.
5. Optional: turn findings into decisions. Bridgespan's Choosing your AI path framework was built on these survey results, and we walked through it in our earlier post. The NTEN AI Resource Hub has more resources to work from.

Pro tip: Staff rated examples and case studies from other nonprofits almost as highly as clear policies, with 57% calling them significantly helpful. If a peer organization will share its AI guidance, start there.

The bottom line

The report costs nothing but a free account. The real investment is time: reading it, then doing the slower work of writing guidance and training your team.

The problem this survey surfaces isn't that nonprofits are slow to try AI. It's that everyday use has outpaced the rules and skills that keep it safe for your staff and the people you serve. The hopeful part is that staff and leaders, asked separately, pointed to the same answer. Your team is probably using AI already. The question is whether they have what they need to use it well.

References

• Two-page summary of the State of Nonprofit AI Adoption and Governance
• State of Nonprofit AI Adoption and Governance: Source article, NTEN, September 10, 2026 (report page; sign in with a free NTEN account to access the full report)
• How Nonprofits Adopt and Govern AI: Insights from a New Report: Coverage and analysis, Nonprofit Quarterly (Coty Poynter), September 16, 2026
• NTEN: a community of nonprofit professionals offering education and research on nonprofit technology
• The Bridgespan Group: a global nonprofit advising social change leaders since 2000
• Choosing your AI path: A framework for nonprofit leaders to make strategic choices: Bridgespan's framework, built on these survey results
• NTEN AI Resource Hub: NTEN's collection of AI resources for nonprofits
• Bridgespan & NTEN Release AI Strategy Framework for Nonprofits: our earlier post on the framework