top of page
Search

Transparency: What Your Donors Deserve to Know About Your AI

Oct 5
5 min read
TL;DR: Most nonprofits now use AI somewhere in their fundraising — but very few tell their donors about it. Transparency isn’t a line buried in a privacy policy. It’s a relationship obligation: your donors handed you their trust along with their money, and they deserve to know when a machine is shaping what they see, how they’re asked, and what happens to their data.

A few months ago, a fundraiser told me — almost in passing — that her organization had started using an AI tool to score donors, to predict who was most likely to give a major gift. I asked whether the donors knew. She paused. “Why would we tell them?” she said. Not defensively. Genuinely. It simply hadn’t occurred to her that it was a thing to tell.

That pause has stayed with me. Because the honest answer — the one I didn’t say out loud fast enough — is that we tell people things not because the law forces us to, but because withholding them would feel like a small betrayal if they found out later. That’s the test I keep returning to. Not “are we required to disclose this?” but “how would this land if a donor read about it somewhere other than from us?”

What does transparency actually mean when we talk about AI?

Transparency is one of the six layers in the AI Equity Framework I work from, and it’s the one people most often shrink down to a checkbox. It isn’t the same as explainability — being able to open up a model and describe how it reached a number. That’s useful, but it’s an engineering problem. Transparency is quieter and more human. It’s telling people the truth about what’s happening, in words they can actually understand, before they have to ask.

In practice it means a donor can know that AI is involved, roughly what it’s doing, and what it’s doing with their information. Not the architecture. The relationship. You are not obligated to teach anyone machine learning. You are obligated not to let them assume a human read their note when a model did.

Why would donors even care?

There’s a quiet assumption underneath the silence — that donors won’t notice, and if they don’t notice, no harm is done. I don’t think that holds. People care about being treated as people rather than as data points. A donor who learns, after the fact, that an algorithm sorted them into a “low potential” bucket and quietly moved them from a phone call to an automated email tends not to be upset about the technology. They’re upset about the demotion they never knew happened.

And here’s the part that surprises people: plenty of donors would be glad to hear it. Told that you use AI to cut down on busywork so your two-person team can spend more time on the mission, many will nod. Transparency lets the people who object opt out, and lets the people who approve feel good about it. Silence takes that choice away from both of them.

But isn’t it safer to just not mention it?

This is the counterargument I hear most, and I want to give it its due, because it isn’t foolish. If we tell donors we use AI, the worry goes, we’ll scare them, or invite scrutiny we’re not ready for, or imply we’ve swapped out the human relationships they value. And there’s something real in that. A clumsy disclosure — one sentence of jargon dropped into a footer — can do more damage than saying nothing.

But “it might be awkward” is not the same as “it’s better hidden.” Most of the discomfort of disclosing turns out to be the discomfort of not yet having a good answer to an obvious question. That’s not a reason to stay quiet. It’s a reason to go get the answer. If you can’t comfortably explain to a donor why you’re using a tool, that’s worth noticing — it may be the tool, not the telling, that’s the problem.

How much is too much to share?

I’m not arguing that every use of AI needs its own billboard. Transparency without proportion just becomes noise, and a donor buried in disclosures is no better informed than one kept in the dark. You don’t need to publish your model weights or send a notice every time software checks a spelling. The standard I use is simpler: does the AI shape how a human being gets treated? If a tool decides who gets a personal call and who gets a form email — then yes, say so. If it’s correcting your grammar, let it be.

The line, roughly, is consequence. The more a system touches a real decision about a real person, the more they deserve to know it’s there. Here’s where I’d start if you’re building that practice from scratch:

  • Name it plainly. Say “we use AI to help us…” in language a donor would use at their kitchen table, not in terms a vendor would use in a pitch deck.

  • Describe what it does, not how it works. Donors want to know the effect on them — that a tool helps prioritize outreach — not the mechanics of the algorithm behind it.

  • Offer a way out. Give people a clear path to ask that their data not be used this way, and make honoring that request genuinely easy, not a maze.

  • Keep a human on consequential calls. Let AI inform decisions about who gets what, but make sure a person signs off before anything that materially changes how a donor is treated.

  • Be honest about the data. Tell donors, in plain terms, whether their information trains, feeds, or leaves the tool — especially if a third-party vendor ever touches it.

Transparency doesn’t cost you the relationship. It’s the thing that lets the relationship survive the technology.

Frequently asked questions

Do we legally have to tell donors we use AI?

In most places there is no single law requiring a nonprofit to disclose general AI use to donors, though data-protection rules such as GDPR and some state privacy laws do create obligations around automated decision-making and personal data. But legality is the floor, not the goal. Transparency with donors is an ethical commitment that usually goes beyond what any statute demands, and treating a disclosure as a compliance task rather than a trust-building one tends to produce the bare minimum, which donors can feel.

Won’t disclosing our AI use make donors trust us less?

The evidence and experience point the other way more often than not. Donors tend to lose trust when they discover something was hidden, not when something is explained to them openly and in advance. A clear, jargon-free disclosure that says what a tool does and why usually reassures people, because it signals that the organization is thinking carefully about their interests. The risk is far greater when a donor learns about your AI use from someone other than you.

Where should AI disclosure actually live?

It should live wherever donors are most likely to look and most affected by the tool, not only in a privacy policy few people read. That can mean a short, plain-language section on your website, a line in donor communications when AI meaningfully shapes the interaction, and a clear answer ready whenever someone asks directly. The aim is that a reasonably curious donor would feel informed rather than managed — disclosure placed to be found, not technically present but practically hidden.

 
 
 

Comments


bottom of page