AI Literacy for Nonprofit Boards: Questions Every Trustee Should Ask
Most nonprofit boards did not plan to become AI overseers. The tools arrived quietly: a grant-writing assistant one program officer started using, a donor-scoring feature switched on inside the CRM, a chatbot added to the website by a vendor during a routine update. By the time AI shows up on a board agenda, it is often already shaping how the organization raises money, talks to supporters and decides who gets served.
Trustees do not need to understand how a language model is trained to govern it well. They need something more ordinary and more useful: the habit of asking the right questions, at the right moments, and noticing when the answers are vague. That is what AI literacy means at board level. It is less about technology and more about fiduciary duty, mission alignment and risk.
This guide offers a practical set of questions for trustees, grouped by the decisions boards actually make. Use them in committee meetings, in conversations with the executive director, or as a checklist before approving a new system or vendor contract.
Why AI Belongs on the Board Agenda
Boards already oversee finances, legal compliance and reputation. AI touches all three. A tool that mishandles donor data creates legal exposure. A model that steers outreach away from certain communities creates reputational and mission risk. A subscription that grows quietly every year becomes a budget line nobody approved.
Governance, Not Operations
The board’s job is not to pick software or write prompts. Staff should own day-to-day decisions. The board’s role is to make sure those decisions follow clear principles, that someone is accountable, and that the organization would be comfortable explaining its choices to donors, beneficiaries and regulators.
The Cost of Silence
When boards stay out of the conversation, a common pattern appears: adoption happens team by team, each with its own tools and assumptions, and no one has a full picture. Problems surface only after something goes wrong. A few well-placed questions early on are far cheaper than a crisis review later.
Questions About Strategy and Mission Fit
Start with purpose. AI should serve the mission, not the other way around. Many organizations adopt tools because peers are doing it or because a vendor offered a discount, without asking what problem they are solving.
What Problem Are We Solving?
Ask staff to describe the specific workflow the tool will change, who does that work today, and what success would look like in six months. If the answer is “efficiency” with no further detail, push for specifics: hours saved on which task, errors reduced in which process, which people freed up to do what instead.
Does This Strengthen or Replace Relationships?
Giving runs on trust between people. A useful test is whether a tool frees staff to spend more time with donors, partners and the communities served, or whether it substitutes automated contact for human contact. Neither is automatically wrong, but the board should know which one it is approving.
- Which human interactions will this tool change or remove?
- How will supporters know when they are dealing with an automated system?
- What would we lose if the tool disappeared tomorrow, and could we cope?
Questions About Data, Privacy and Consent
AI tools are only as trustworthy as the data they use and the way they handle it. For boards, this is where legal duty and donor trust meet. The earlier discussion of personalization and privacy in AI-powered fundraising shows how quickly helpful targeting can start to feel intrusive.
Where Does Our Data Go?
Trustees should know whether donor, beneficiary or staff information is sent to an outside provider, whether that provider can use it to train its own models, and where it is stored. Contracts should answer these questions in writing. If staff cannot find the answer, that is itself an answer.
Did People Agree to This Use?
Donors gave their information to support a cause, not necessarily to be scored or profiled. Beneficiaries often share sensitive details because they need help, not because they consented to analysis. Ask whether current privacy notices cover the new use, and whether the organization meets the rules that apply to it, such as GDPR for European supporters or state privacy laws in the US.
Questions About Risk, Bias and Accountability
Every AI system makes mistakes. The governance question is whether the organization will notice them, who will fix them and how harm will be addressed.
Who Is Accountable When It Goes Wrong?
Name a person, not a department. If a donor-scoring model wrongly deprioritizes a loyal supporter, or an eligibility tool screens out a family that should qualify, someone must own the review and the response. Boards should expect a clear escalation path and a log of significant incidents.
Have We Tested for Unfair Outcomes?
Models trained on past data can repeat past patterns, including who was overlooked. Algorithmic bias is not an abstract concern for organizations that distribute services or grants. Ask whether results have been compared across groups, and whether anyone outside the project team has looked at them.
A board does not need to understand the model. It needs to know who is watching it, what they are looking for and what happens when they find a problem.
Questions to Ask Before Approving a Vendor
Many AI decisions are really procurement decisions. Before signing, trustees on the finance or audit committee can use a short set of questions to separate careful vendors from careless ones.
| Question | Why It Matters | Warning Sign |
|---|---|---|
| Can we export our data and leave? | Avoids lock-in and protects records | No export format or high exit fees |
| Is our data used to train your models? | Protects donor and beneficiary privacy | Opt-out buried in terms, or no answer |
| How do you explain a specific output? | Lets staff check and justify decisions | “The model is proprietary” |
| What does pricing look like in year three? | Prevents budget creep | Introductory rates with no cap |
| Who do we call when something breaks? | Ensures real support for small teams | Support only through a chatbot |
Total Cost, Not Sticker Price
The license fee is rarely the full cost. Staff training, data cleanup, integration with existing systems and ongoing review all take time. Ask for an honest estimate of internal hours, not just the vendor invoice.
Turning Questions Into a Standing Board Habit
Good questions depend on a baseline of understanding. Boards can build it without turning trustees into technologists.
Practical Steps for the Next Year
- Ask management for a simple inventory of AI tools in use, including features inside existing software.
- Schedule one board session where staff demonstrate a tool on real (anonymized) work.
- Adopt a written policy; the framework in building an AI policy for your nonprofit is a reasonable starting point.
- Assign AI oversight to an existing committee rather than creating a new one.
- Review the inventory and any incidents once a year, alongside the audit.
Recruiting for Gaps
When a board seat opens, consider whether someone with data, privacy or technology governance experience would help. The goal is not a resident expert who answers every question, but a trustee who is comfortable asking follow-ups and reading a vendor contract critically.
Making Oversight Routine
A single good discussion is not oversight. What matters is repetition: the same core questions asked whenever a significant tool is proposed, renewed or expanded. Some boards add a short AI section to the template used for new program and budget proposals, so the questions travel with the decision rather than relying on memory.
Imagine a mid-sized community foundation where the grants committee chair asks three things every time a new system comes up: what problem it solves, what data it touches and who is accountable. Within a year, staff begin answering those questions before they are asked. That is what board-level AI literacy looks like in practice, not technical fluency, but a culture where careful adoption is the default.
Trustees who ask steady, specific questions protect more than the budget. They protect the trust donors place in the organization and the dignity of the people it serves, which is the whole point of the work.
