We Don't Have This Figured Out Either: AI, Donor Data, and the Questions Every Nonprofit Should Be Asking

Somewhere on your team, someone has probably already pasted donor notes into ChatGPT to draft a thank-you letter faster. Someone else may have asked an AI tool to help analyze giving patterns, or summarize a board packet. Chances are, no one decided this was okay. It just started happening, the way most technology adoption does — quietly, individually, without a conversation.

We're not writing this because we've solved it. We're writing it because we started asking the questions out loud, and we think every nonprofit should be asking them too.

The donor data question

Names, gift histories, giving capacity, personal notes from a major donor visit — this is some of the most sensitive information a nonprofit holds, entrusted by people who gave because they trusted you with it. AI tools are genuinely useful for drafting, brainstorming, and strategy. They're a different animal entirely when donor-identified data gets typed into them. Most consumer AI tools' data-handling policies weren't written with donor trust in mind, and most staff using them have never been told where the line is — because most organizations haven't drawn one.

The environmental question

These tools have a real footprint — the energy and water used to run the data centers behind them, at a scale that's still being measured and debated. For organizations whose missions are about community, sustainability, or stewardship of resources, this isn't a footnote. It's worth sitting with the tension: a tool that saves staff time may also carry costs that don't show up on your budget line.

The trust question

This year alone brought a steady stream of stories about AI systems behaving in ways their own developers didn't fully anticipate — guardrails not holding, data ending up somewhere it shouldn't, tools acting outside the boundaries they were supposedly built with. (We'd encourage you to look into the specifics yourself rather than take our word for it — this is a fast-moving story.) The point isn't that AI is untrustworthy. It's that "the company built in safeguards" isn't the same as "we understand exactly what those safeguards do and don't cover." Nonprofits, more than most organizations, can't afford to find that out the hard way with donor trust on the line.

Questions worth bringing to your next team or board meeting

- Where is AI already being used in our organization — officially or not?

- What counts as sensitive here? (Hint: it's more than just Social Security numbers.)

- Is there anything we'd never want typed into an AI tool, even by accident?

- Who decides what's okay, and where is that decision written down — or is it just assumed?

- What would it cost us — in trust, not just dollars — if we got this wrong?

We don't have a tidy answer to hand you. What we have is a habit of asking these questions before we act, and a genuine belief that the nonprofits who start asking now — even without perfect answers — will be in a far better position than the ones who wait until something goes wrong.

If your team hasn't had this conversation yet, consider this your nudge.