As charities race to adopt artificial intelligence, “the single most important step” should be a clear framework on how it is used. But only 10% of Canadian non-profits have formal policies.
Just over three years ago, before the launch of ChatGPT in late 2022, no one was using artificial intelligence daily. Talking about it? Yes. Mulling where it might go? Certainly. But no one – literally, no one – was talking about using AI to write their grant reports, or to draft donor emails, or to create fundraising campaigns. All of this was but an apple in a tech giant’s eye.
But as hundreds of professionals in the fundraising space gathered at the recent Canadian Association of Gift Planners conference in Winnipeg in April, keynote speaker Nathan Chappell, chief AI officer at Virtuous, asked the room who uses AI in their work. By his estimate, 97% of attendees put their hands up. That tracks with recent research from the 2026 Nonprofit AI Adoption Report, he said. “We found out that 92% are using AI,” he said of the global survey of close to 350 organizations in the charitable and non-profit space. “But 80% are using it individually on an ad hoc basis.”
That reality – rapid adoption, but little strategy or policy – is true in Canadian non-profits, too. “While most nonprofits experiment with AI, few have developed internal policies to guide its use,” reads a January report from Imagine Canada, which found that only 10% of Canadian non-profits have formal AI policies. Not only that: “64% of nonprofits using AI have no policies and aren’t developing any.” And that, experts say, creates risk. Risk for organizations’ data security, but also for their reputations with donors.
“The current extensive charity use of AI without visible, effective governance, clear human control, and honest communication creates a risk of our own making,” an April report from the Charity Excellence Framework in the United Kingdom concludes. That risk has real consequences when it comes to trust – and charities and non-profits can only exist with trust. The Charity Excellence Framework warns that the lack of transparency is “creating a growing risk to public trust unless governance and transparency catch up.”
One of the most striking findings is that 83% of people – especially donors – want transparency about how charities use AI.
Charity Insights Canada Project research
Similar risks exist in Canada. Research from Carleton University’s Charity Insights Canada Project recently sought to find out how donors feel about AI use by the charities and non-profits they give money to. “One of the most striking findings is that 83% of people – especially donors – want transparency about how charities use AI,” the survey found.
And the trouble right now, in this moment of rapid AI adoption, much of which is ad hoc, is that even organizations themselves don’t know how they use AI. “The conversation is, as an organization, can I confidently identify where we’re using AI?” said Michelle Gazze, community lead of Neralake, which demonstrated an AI tool specifically for non-profits for strategic planning processes and capacity-building at the CAGP conference. With AI, Gazze added, “innovation expands, but also the responsibility does as well.” The tools themselves “are never, ever going to be held responsible,” she added. “It’s the organization, and the people who are using them that are.”
And that leads many who are closely watching the rollout of AI in the non-profit space to the same conclusion: transparency and clear policies are key. “I don’t think governance gets enough attention,” Chappell says in an interview. “It’s not the sexiest part of it.”
It’s important to understand that when we start to use these tools, oversight becomes really, really necessary.
Matt Dirks, Neralake
“It’s important to understand that when we start to use these tools, oversight becomes really, really necessary,” said Matt Dirks, managing partner with Neralake, at a presentation at the CAGP conference on ethical AI use. “It’s so important to ensure that we’re not just protecting ourselves and our work, but we’re also protecting that trust that we have with our stakeholders and the folks that trust us.”
And that is going to become only more important as AI evolves. “The technology continues to do increasingly impressive things. Having an evaluation matrix on when to go or not go or go with caution is imperative,” Chappel says, adding that that’s true for both organizations and humans. “We all have to weigh the short-term and long-term consequences of decisions that we make – just the same way we decide if we’re going to eat ice cream for breakfast.”
AI without guardrails can challenge trust in the charitable sector
In many ways, trust is the bread and butter of the charitable and non-profit sector. “The entire sector is sustained by the faith donors have in charities,” Luke Johnson of the L. Johnson Law Group said at the conference. “Canadians still have a very high level of trust in charities,” he said. “That high level of trust is based on the last 50 years of successful work.”
A 2025 public opinion poll from Imagine Canada confirms the sector’s good reputation. In it, 82% of respondents said they trust the charitable and non-profit sector (that’s compared to the level of trust in government, which clocked in at 48%).
“The charitable sector itself has a large degree of public trust. It has more trust than most industries will ever, ever, ever earn,” Gazze said. “And it’s really important that you clarify how you use AI . . . because not having that clear direction could put a lot of things at risk for your organization.” And, she added, “the biggest thing at risk is trust, because your teams have access to a lot of information most industries don’t. You have donor financial information. You have clients’ immigration status; you have housing status. You have all this information that communities really trust you with.”
The biggest thing at risk is trust, because your teams have access to a lot of information most industries don’t.
Michelle Gazze, Neralake
It’s not just data concerns that organizations need to be thinking about. There are also real concerns about the ethics and sustainability of using AI, many of which are shared by some donors, Dirks said.
Indeed, huge new technologies – automobiles, oil, plastics – have been rolled out in the past with much fanfare and speed . . . and seemingly little thought to the myriad downsides of their widespread use. As AI ramps up, concerns about its future impacts are real, with the International Energy Agency predicting that “electricity consumption from data centres is set to double by 2030.” According to the agency, an individual server rack within a data centre is the size of a refrigerator, but by next year one rack alone could use as much power as 65 households.
“The AI build-out is mind-blowingly resource-intensive: really, really, really big power plants being built in places that are not great for water security, not great for the communities that they’re in,” Dirks said. “So it is really important that we don’t separate ourselves from that reality. There are trade-offs anytime that we decide to use [an AI] tool.”
In the end, for charities and non-profits, clear communication about AI use has emerged as a key component of any AI strategy – and is key to building, and maintaining, public trust.
The ‘single most important step’ non-profits can take
According to Chappell, creating an AI governance policy is “the single most important step” a non-profit can take to move from reactive AI use to strategic impact. “It actually is not that hard,” he says. He recommends looking at an AI framework (he points to those available on fundraising.ai) and the organizations’ own values. “Build a policy from those two documents,” he says. “Governance is essentially the starting point that serves as basically a traffic signal for an organization to either move forward, not move forward, or move forward with caution.”
But, he adds, that’s just the beginning. “That’s step one, and step two is what we consider the acceptable use policy, or acceptable use guidelines . . . which are where the rubber hits the road. That’s where organizations actually specifically state which tools that they use for what purposes.”
The latter, he says, needs to be a living document. Dirks agrees, adding that these guidelines need to be updated constantly and need to be “very pragmatic and accessible for team members, ensuring that it’s always in plain language.” It’s key, he adds, that non-profits have “a demonstrated knowledge of where the tools are actually being used in the workplace.”
There are also huge differences in the implications of AI use at different levels, Dirks says. “Not all use cases are the same. There’s big differences in what the consequences are of a public-facing or stakeholder-facing communication [versus] working off the side of my desk on drafting my thoughts coming out of a meeting for my own self.”
That means non-profits need to be thinking about a tiered approach to how they use AI – and how they communicate its use externally. “Organizations that are going to be successful, with implementing and working around the trust and the trade-offs, are going to be ones that are leading from their values and being very transparent about what they’re trying,” Dirks says. “It’s not something that you can just set and forget. It’s an ongoing conversation.” He adds, “It is a tough thing to navigate, and will continue to be a tough thing.”
The greatest risk with all this is not just to the organization itself, but to all adjacent organizations.
Nathan Chappell, Virtuous
And, as Chappell points out, the risks don’t stop at the individual organization. “The greatest risk with all this is not just to the organization itself, but to all adjacent organizations,” he says. “Non-profits are unique in the sense that they’re built and owned by the public . . . And so therefore, they’re all kind of connected.”
That means that one catastrophic misstep has the potential to affect not just the individual organization, but the reputation of the whole sector.
Chappell recently wrote about a hypothetical example he dubbed “counterfeit generosity” or “autonomous fundraising” – the idea of, say, an AI-generated voice making fundraising calls. He paints the picture of a warm voice calling you up, one that knows all the details about your giving history and just the right things to say. Then, as the realization dawns that the whole thing was in fact AI, he writes, comes the feeling of being duped.
The reputational damage to the whole sector is potentially huge, if donors lose faith in the human connection at the heart of fundraising. “Just because we can do something does not mean we should, and the difference, in the end, comes down to whether anyone took the time to ask the question at all,” Chappell writes.
Another AI risk: amplifying scams
At the same time charities open themselves up to risks from using AI without clear policy, Johnson cautioned that charities are also at risk of scams like anyone else – and “tech is massively multiplying the reach of scams.”
“People are deploying AI to do this faster and cheaper,” he added, pointing to AI live video capacity as an example.
Scammers have successfully targeted Canadian organizations in the sector, creating headaches and, in some cases, serious fallout. Nearly 60% of Canadian charities responding to a survey from Charity Insights Canada in June 2024 reported they had been the subject of phishing or email scams, and another 41% reported fake invoices and billing scams.
As the Government of Canada put it in a March warning about AI impersonators of government employees, “Artificial intelligence (AI) is giving fraudsters powerful new tools to deceive.” The warning specifically mentioned the risks of voice cloning, AI-generated website spoofs, and AI-generated messages.
That has led many to warn that organizations need to be thinking not just about the risks to public trust when they deploy AI tools internally, but about the risks posed to them by AI-driven scams.
Humans are still central to charitable work – and fundraising
There’s a dichotomy at the heart of narratives around AI. On one hand, headlines boast that “AI is driving huge productivity gains.” On the other, there’s a heated debate about the costs to the environment, to jobs, to our brains.
“AI is a really hard topic, because there’s everything from ‘AI is going to destroy the planet’ to AI FOMO,” Chappell says. But he doesn’t mince words about where he lands on the importance of AI in non-profits. “The way that we think about our work is fundamentally changed once you know that reality will divide those that use AI versus those that don’t,” he says. “AI will not replace people or organizations, but people who use AI, organizations that use AI, will replace those that don’t.”
AI will not replace people or organizations, but people who use AI, organizations that use AI, will replace those that don’t.
Nathan Chappell
Still, at every panel on AI at the CAGP conference, one takeaway was the same: humans still need to be at the heart of non-profits’ work. And that becomes more true with the rise of AI – not less.
“We’re in this world where we have so much information, so much being thrown in front of us on a daily basis,” Doug Darling of Tripwire Media Group said, adding that when we “pour AI all over that,” the end result is an overwhelming stream of AI slop.
That’s a key challenge for charities and non-profits that are trying to cut through the noise and connect with potential donors on a human level. “All these things that were special are becoming table stakes,” Darling said. That makes authenticity – and clear communication of that authenticity – even more key. “Story is your superpower that breaks through AI clutter,” he said. “You need to know who you are and you need to stick to that.”
The sources included in this article presented at the CAGP conference in April in Winnipeg.
