“Nonprofits and NGOs are adopting (generative) artificial intelligence technologies faster than they’re designing governance and training for their teams.” Thus begins the 2026 State of Nonprofit AI: Adoption and Governance Report.
Researched and penned in partnership by NTEN and The Bridgespan Group, the report sets out to explain “how nonprofit and social sector organizations are adopting and using AI tools and technologies—from generative AI tools to AI-powered features embedded in everyday software.”
But as public perception of AI is increasingly scrutinized—from concerns over doomsday scenarios that arise from a lack of guardrails to the environmental costs of data centers—where does AI factor into the nonprofit sector?
According to information gathered during a survey from late April to mid-June, of the 917 respondents, 98 percent reported using AI in some capacity. And 61 percent use AI in an official capacity, which the report considers “organized pilots (33%), team-wide use (24%), or fully integrated use (4%).” Perhaps surprisingly, only 2 percent of respondents don’t use AI at all.
“…57 percent of nonprofit executives report no dedicated AI budget, 37 percent don’t currently train staff on how to use the AI tools available to them, and 58 percent report that no AI roadmap exists.”
With the documented AI use by respondents, the report’s authors highlight three main trends: interest in AI has outpaced the infrastructure required to support it; staff and executives are experiencing AI adoption very differently; and the sector already agrees on the fix: clearer guidance and real training. The report also provides a framework for nonprofit leaders that can help inform decisions about AI implementation within their organizations.
The Implementation-Governance Gap
When it comes to AI use, implementation is outpacing governance at most organizations. The report offers five categories (AI-related budget, staff training for AI tools, infrastructure such as cloud storage and computing resources, data organization, and staff knowledge of responsible use) that respondents weighed in on. Budget is by far the clearest gap, but infrastructure, strategy, and training all lag too.
According to the report, 57 percent of nonprofit executives report no dedicated AI budget, 37 percent don’t currently train staff on how to use the AI tools available to them, and 58 percent report that no AI roadmap exists. Only 8 percent have one in place. Meanwhile, less than half executives surveyed have written guidance in place for what data can be used with AI tool (40 percent) or how to responsibly use AI (38 percent).
“With formal policy this underdeveloped, a meaningful amount of AI use across the sector—including among executives—falls outside clearly defined guidance by default,” the authors write. This presents a new question: If AI use is widespread but fewer than half have written guidance on how to use it responsibly, what does that say about where the sector’s priorities have landed?
Executive Decisions, Staff Impact
As for who makes the decisions within nonprofit organizations regarding AI implementation, it is perhaps no surprise that the tendency is for it to happen at the top and trickle down. That results in very different experiences with how AI impacts day-to-day operations.
“AI use has already become a routine part of the workday for most of the nonprofit sector,” the authors write. “45% of respondents report using AI for work at least daily, and another 29% use it regularly, about once a week.”
At a glance, executives tend to turn to AI more readily than staff and are more comfortable relying on it as a tool. In response to how often AI is used in day-to-day work, 50 percent of executives use AI daily or more while only 38 percent of staff do. Executives are less likely to shy away from AI use, with only 3 percent reporting that they never use it, compared to 13 percent of staff. Perhaps more surprising are the respondents who use AI less than once a month (8 percent) or never use it in their day-to-day work (7 percent).
“Staff more often describe adoption as something happening ‘to and around them, driven by external pressures or day-to-day work needs rather than being intentional, collaborative, and planned.’”
The degree of use mirrors the comfort level, with 67 percent of executives reporting that they’re comfortable using AI tools compared to only 55 percent of staff. Meanwhile, staff are nearly twice as likely to report being very uncomfortable (19 percent versus 10 percent).
Sign up for our free newsletters
Subscribe to NPQ's newsletters to have our top stories delivered directly to your inbox.
By signing up, you agree to our privacy policy and terms of use, and to receive messages from NPQ and our partners.
In terms of the reasons behind decisions to adopt AI, the report indicates that executives tend to offer deliberate strategy around AI use, while staff more often describe adoption as something happening “to and around them, driven by external pressures or day-to-day work needs rather than being intentional, collaborative, and planned.”
In other words, executives typically make decisions and are responsible for AI implementation at their respective organizations (63 percent). One-fourth of staff (25 percent) indicated there was an internal AI council or ethics committee that shaped decision-making around its use.
But there is an outlier here, with only 1 percent reporting a collective or all-staff decision around AI use.
More Guardrails and Ethical Training
To bring it full circle, we have to look back at the implementation-governance gap, which highlights the need for guardrails and ethical training for AI tools. The authors note that 53 percent of all respondents report shadow use—informal, unofficial AI use outside of organizational guidance—and counterintuitively, executives engage in shadow use at a higher rate than staff (57 percent versus 49 percent).
Nonprofits have the opportunity to create ethical AI policies, but that starts at the top and trickles down—in much the same way as decision-making around an organization’s implementation of AI. Once that choice is made, thus begins the Sisyphean struggle to ensure responsible usage of a nascent (and controversial) technology.
As with all new technology implemented within an organization, it doesn’t come without its issues. One of the top concerns across the board for executives (55 percent citing it as a major barrier) and staff is data privacy (56 percent citing it as significant)—and 66 percent of staff cite environmental impact as a major concern, which was the single highest response across any concern category measured for staff. Meanwhile, 47 percent of executives cite the widening gap between populations with and without access to AI-enabled services as their top field-level risk.
“Is the low rate of reported harm evidence that the sector is managing AI responsibly or evidence that most organizations don’t yet have the systems in place to know when something has gone wrong?”
Although the concerns around AI outlined by respondents are consistent—data privacy, environmental impact, bias, community trust—reported harm is rare, with only 6 percent of respondents reporting an AI-related incident in the past 12 months.
That said, as the report notes: “This question captures realized harm over the last 12 months, not the sector’s current exposure to risk—harm may be going undetected or undisclosed rather than genuinely not occurring.” This then raises a key issue: That gap may say more about the absence of incident-tracking than the absence of harm itself because its implementation is so new and guidance undercooked. Is the low rate of reported harm evidence that the sector is managing AI responsibly, or is it evidence that most organizations don’t yet have systems in place to know when something has gone wrong?
The Fix for AI Governance
The picture isn’t entirely bleak within the nonprofit sector’s implementation of AI. There is—and likely always will be—some degree of divide around why to use AI in terms of the ethical and environmental trade-offs, and that might be addressed through either a careful cost-benefit analysis or deliberate inaction.
The report is clear that guidance and actual training are necessary. As it stands now, only 32 percent of staff have received formal AI training—even at well-resourced organizations, nearly 60 percent of staff are untrained. At the same time, the report indicates that 49 percent of executives anticipate AI spending will grow moderately or significantly in 2026. Despite this, AI governance hasn’t kept pace with what’s already been spent. To that end, the authors offer a framework—Bridgespan’s Choosing Your AI Path—as a practical starting point for nonprofit leaders to better understand how to think more strategically about technology.
The question then becomes whether organizations will prioritize it before the gap between adoption and governance widens further. Should the moment be capitalized on and the gap closed, the nonprofit sector has an opportunity to, as Jean Westrick wrote in her article on building AI for the public good, “ensure AI serves the public good, advances equity, and strengthens the nonprofit sector rather than exacerbate existing disparities.”
If the solutions are known, then it is merely a matter committing dollars to guidance and training that makes the implementation of AI feel less abstract and more direct. When that happens, the nonprofit sector stands a better chance at ensuring AI does, in fact, serve the public good.
