What should we tell members who ask how our association uses AI?

The short answer: tell them what AI is used for, what it is not used for, who is accountable for it, and how to ask a question. Say only what you have checked. Members are not asking you to avoid AI; they are asking you to be straight about it.

The question usually arrives sideways. A member emails the chief staff executive after seeing an unusual chatbot answer. A board member asks at a meeting whether the newsletter was written by a machine. Another member wants to know if their data went into a model. None of these is hostile. All of them are easier to answer when the answer already exists.

What members and the public are saying they want

The evidence is mostly from surveys of the public and of people who support charities, rather than association members specifically, so treat it as direction, not a census.

• A 2026 Candid report found 76% of respondents agreed nonprofits should fully disclose when and how they use AI, and more people said nonprofit AI use would lower their trust than raise it.
• A survey of 1,006 U.S. charity supporters, reported by the Chronicle of Philanthropy, found 86% rated disclosure of AI use as very or somewhat important. About 31% said they would be less likely to support charities using AI, 34% said it would make no difference, and 9% said they would be more likely.
• In ISACA's 2026 poll, only around 18% of organizations both require and enforce disclosure of AI use. We wrote about that gap here.

The pattern is consistent: people are uneasy about AI, and they are much less uneasy about an organization that explains itself.

What a good answer contains

A useful statement is short, specific and boring. It covers four things.

1. Where AI is used. Name the functions, not the products: drafting, member-service chat, event scheduling, exam development, fraud screening.
2. Where it is not used. If a person makes every decision about a membership, a credential or a grant, say so, and only if it is true.
3. Who is accountable. A named role, not "our team".
4. How to ask. An address or form, and a promise about how quickly someone will reply.

A sample statement

Adapt this, and change anything that is not true of your organization.

We use AI tools for [drafting communications, answering common member questions, and scheduling]. A staff member reviews anything AI helps produce before it reaches members. We do not use AI to make decisions about [membership, credentials, or awards]. [Role] is responsible for how we use AI. Questions are welcome at [address], and we aim to respond within [number] business days.

The brackets matter. A statement with a promise your organization cannot keep is worse than no statement.

What to verify before you publish

The hard part is not the wording. It is learning whether the sentences are true. Four checks cover most of the risk:

• Vendor-embedded AI. Your association management system, email platform, event tools and exam vendors may have added AI features you did not choose. We wrote about this here.
• Staff use. People use AI tools the organization never approved, usually with good intentions. Ask, without blame.
• Data. Find out whether member data is sent to or used to train any AI service, and get the vendor's answer in writing.
• What your member-facing assistants actually say. Test them with the questions members really ask, and keep the word-for-word results. A statement that says "a person reviews our answers" is only as good as the last time anyone checked, so this is a test to repeat, not run once. That is what CimAssure does: quarterly, from the outside, the way a member would meet your assistant.

Common questions

Do we have to disclose? That depends on your jurisdictions and the decisions involved, and it is a question for your counsel. Where no rule applies, disclosure is still the expectation members and the public are voicing.

Should we wait until we have a full AI policy? No. A short, accurate statement now beats a perfect policy later. Update it as practice changes.

What if we find something we would rather not disclose? Then you have found the thing to fix first. Do not publish a statement that the facts do not support.

Who should sign off on the statement? The chief staff executive, with board awareness. It is a statement of the organization's practice, not a marketing line.

How do we keep the statement true after we publish it? Treat it as a claim that needs evidence. Assign the accountable role, review the statement whenever a vendor or a workflow changes, and keep dated test results for your member-facing assistants. Quarterly CimAssure testing produces that record, and the report goes to you, not the public.

What we do not know

We have no good data on how association members specifically respond to disclosure. The research above is about the public and charity supporters, and the Candid figures reach us through a secondary report. They point the same way, but nobody should claim to know what your members think.

Where to start

Our AI Trust Readiness Scorecard takes about five minutes, with no sign-up to see your result. If you want an independent view of where AI is running in your organization before you say anything publicly, the AI Assessment is a fixed-price, three-week review. Your report stays confidential to you, and you decide what, if anything, to publish. Once your statement is out, CimAssure keeps it honest with ongoing quarterly testing of your member-facing assistants. The framework is published.