Fundraisers want AI, but don't know how to use it responsibly

AI is already woven into the daily work of fundraisers. The rules on how to use it lag far behind. That is the common thread through two studies published this summer. In France, France générosités surveyed 375 people at organisations that rely on public giving. In the UK, Blackbaud surveyed 380 fundraisers for the ninth edition of its Status of UK Fundraising.

Plenty of use, little structure

The French figures are clear. 86 percent use AI at work. But only 39 percent do so within a framework their organisation has set. The other 47 percent use AI on their own initiative, with no guidance on what data may go into the tool. This is what the report calls shadow AI : AI use out of sight of leadership.

This is not the occasional experiment. Almost three in ten use AI every day, more than eight in ten at least once a week. Mostly to write text (88 percent) and to look up information (69 percent). For fundraising itself, designing and planning campaigns, only 15 percent turn to AI.

The UK report shows the same movement. In 2024, 43 percent still said they did not use AI. By early 2026, that had dropped to 9 percent.

Where it stalls: training and governance, not money

The barriers appear at two moments: at the start and during rollout.

At the start, the French respondents point to two problems above all. One: people do not understand the tools well enough to put them to use (58 percent). They do not know which tool does what, or how to connect it to their own work. Two: the organisation's leadership knows too little about AI itself (53 percent), so no one at the top decides whether and how the organisation uses it. Lack of money matters far less. It comes only fifth, cited by a little over a quarter of respondents.

In organisations already under way, the problem shifts. Three in four get stuck on the same point: there is no internal framework. In practice that means no rules on which tools are allowed, what data may go in, and who is responsible for the result. Half also run into existing tools that do not meet their internal rules (52 percent).

Often there is also no one steering. In most organisations AI lands on the IT department's plate (44 percent). Just 27 percent have set up a dedicated team, and 13 percent have no one internally taking charge. Training lags too: a third of organisations (32 percent) offer no structured way to upskill their staff.

On top of that, a fault line runs between management and the shop floor. The higher someone sits, the more positively they rate their organisation's approach to AI. In large organisations, those with more than 250 staff, 56 percent name the gap in understanding between management and operational teams as a problem, against 38 percent across the whole panel. Management thinks there is a policy. On the shop floor, people work on their own, and that is where most shadow AI sits.

The UK points the same way. The two biggest obstacles there are ethical doubts (61 percent) and a lack of training (59 percent). Budget and resistance to change come at the bottom. Fundraisers want to use AI, but do not know how to do it responsibly.

What fundraisers ask for themselves

Asked what would help, they put guidance on ethical and responsible use at the top (65 percent), followed by concrete examples that worked in their own sector (56 percent). Training, templates and funding come after that.

The concerns are about reliability. 81 percent worry about misinformation, 78 percent about inaccurate output, 75 percent about data leaks. Barely 17 percent worry about their job.

Two testimonies from the research make clear why rules are needed. The first is about grant applications. With AI, a single staff member can write far more applications in the same time than before. The result: trusts and foundations receive so many applications that they can no longer process them. Some have paused or thoroughly reworked their application process because they cannot handle the volume.

The second testimony is about the bond with the donor. In fundraising built on a personal relationship, an organisation loses exactly that personal knowledge when it lets a tool estimate who can give a lot. Someone who truly knows a person knows things no model pulls from the data.

Why data and rules come first

Still, this is not a story about technology alone. It is about data and rules. One UK figure makes that clear: only 9 percent credit growth in income to better use of technology. At the same time, fewer than one in ten say their systems connect well to each other. The two are linked. AI that works with scattered and poorly kept donor data gives poor results back. The quality of the output depends on the quality of the data you put in.

For anyone working with a CRM like Salesforce, an AI policy therefore does not start with the choice of a tool, but with three questions. Which data is in order? Who may use it? And who checks what the AI produces before it goes to a donor? Rules like these do not slow use down. They make sure the use that is already there happens with someone watching.

France générosités is taking this direction further itself. The second part of its study, expected in October, deals entirely with control, ethics and responsible use. Together with Don en Confiance, it has already published a guide to build such AI governance step by step.

The question for your organisation is not whether your people use AI. They already do. The question is whether your data is in order and whether clear rules are in place, so that this use moves you forward instead of causing problems.

Sources: France générosités, "L'usage de l'IA au sein des OSBL, Volet 1" (June 2026, 375 respondents); Blackbaud Europe, "The Status of UK Fundraising 2026" (380 respondents); coverage via the European Fundraising Association ( FR, UK).