Five Questions Your Board Will Ask About AI

Sooner or later, someone is going to ask, "So an AI is writing our thank-you letters?" It might be a board member at a committee meeting, a CEO reviewing operations, a fellow staff member in the break room, a long-time donor who heard something secondhand, or even a member of the news media.

There is a legitimate governance interest in three areas where AI now operates under the Human-Centered AI Framework: donor data, organizational communications, and strategic decisions. The question deserves a thoughtful answer, because how we handle it determines whether the organization’s leadership becomes a partner in this work or an obstacle to it.

Treating AI adoption as a purely technical decision that needs no organizational conversation is one kind of mistake. Letting board anxiety derail an approach that will deepen donor relationships and raise more money is the other.

The questions cluster into five categories. Each one deserves a direct answer rather than a defensive response.

Donor Trust

The worry is that donors will feel deceived if they learn a thank-you note or appeal letter was drafted by AI. The answer is to explain what AI does in the Human-Centered AI Framework. The AI drafts, and the human development professional reviews, edits, approves, and sends. Every communication that reaches a donor has been read by a person who knows that donor. The AI is a drafting tool, the way a speechwriter is a drafting tool for an executive who still owns the message. We still own every word that reaches our donors.

A board member might follow up with, "But is it really from you?" The answer is yes. You reviewed it, you edited it, you decided to send it. You are the author in every sense that counts.

Data Privacy

Data privacy is a governance question, and the kind of question a board should be asking. Where does donor information go when AI tools are in the picture? The specifics depend on the platform and configuration in play. A reasonable explanation for organizational leadership runs roughly along these lines:

Our donor data is part of a cloud-based platform with SOC 2 compliance, encryption in transit and at rest, and enterprise security practices. The AI capabilities are built into the platform rather than a separate service where data gets sent externally. The data policies governing the AI features are clear. Donor information is not used to train future models.

The details vary by platform. The important point is that the answer should be specific, written down, and adopted as policy before anyone asks the question.

Authenticity

Authenticity is the philosophical version of the trust question. If AI helps draft communications, are the relationships still real? It gets at something we intuit correctly: relationships built on deception are empty.

But using a tool to prepare for a relationship is preparation, the same kind of preparation that happens when reviewing a donor’s record before a meeting. When the AI synthesizes three years of conversation notes into a briefing so we walk into a donor visit fully informed, that is the same preparation done more thoroughly. The human relationship happens in the meeting, the phone call, the handwritten addition to the note. AI makes those moments better by ensuring we show up prepared.

Staff Displacement

Some board members will wonder whether AI means the organization needs fewer development staff. I think it is the opposite. AI magnifies the effectiveness of development staff so much that the staff deserves the label of net revenue center, not a cost to be trimmed.

What AI changes is what we spend our time on, not whether we are needed. The Monday morning review, the donor visits, the event planning, the personal phone calls, the strategic thinking about which donors to prioritize and why, is the human work that produces the dollars. AI hands us back the hours that were going to drafting, data analysis, scheduling follow-ups, and administrative tracking, so we can put them into the human work. For organizations with small development teams, AI gives the first staff member the capacity that used to require a second.

I’m convinced that the rise in fundraising success will justify increasing development staff rather than decreasing it.

Accuracy

Anyone who has used AI knows that these tools sometimes get things wrong. They hallucinate facts, confuse details, and produce confident-sounding text that turns out to be bunk.

That is exactly why the framework never permits sending anything to a donor without a human review first. Every thank-you note, every appeal letter, every donor briefing, and every board report goes through the human development professional who knows the donors and can catch errors before they reach anyone. The Monday morning review is the quality control layer that makes the whole framework trustworthy. AI mistakes are caught the same way any staff member’s mistakes would be caught — by a second pair of eyes with institutional knowledge.

Who Adopts the Policy?

Behind those five questions is a sixth one: whose decision is it, really, to set the AI policy? The answer depends on how the organization is governed, and three governance models are common in American nonprofits today.

Under traditional governance, the dominant model in American nonprofits, the board governs and participates actively in organizational life, including fundraising. The development director brings the AI governance question to the board directly, presenting the operational reality and the proposed policy together. The board deliberates, adopts the policy, and authorizes the work.

Under Policy Governance, the model developed by John Carver, the board sets policy boundaries called Executive Limitations, and the CEO runs the development program within those boundaries. An Executive Limitation might address donor data protection, organizational transparency, ethical standards for communications, or responsible use of technology, and the CEO determines the specific AI policies that satisfy those boundaries. The board’s oversight is higher-level rather than operational.

Under the Strong CEO model, sometimes called the Empowered Executive model, the CEO and the development team lead fundraising strategy and execution. Joanne Oppelt and Jimmy LaRose’s book Major Gifts Ramp-Up (NANOE, 2025) makes a thought-provoking case for this approach. For AI governance, the implication is straightforward. The CEO establishes AI policy as part of operational leadership, and the board’s role is fiduciary oversight, ensuring the organization’s policies protect donors and serve the mission.

The five answers above hold up under any of these three models. What differs is who adopts the policy, how the conversation is structured, and how quickly the organization can move. The work itself comes down to clear policies on donor data, AI disclosure, and ethical boundaries. Those policies need to exist before someone asks one of those five questions and finds we do not have an answer ready.

In my book, Holding Fire, I walk through these governance models in detail, including the AI use policy itself and the forward-looking development report that shifts the boardroom conversation from "how much did we raise?" to "where are we headed?"

Next week we turn to the ethical questions at the heart of all of this. When does knowing donors well become knowing too much? The "honored or handled?" test is the principle that keeps the human center where it belongs.