Key takeaways
- The IRS says exempt organizations generally need not publicly disclose contributor names and addresses, except private foundations and 527 political organizations.
- California's reasonable security statute defines covered businesses to include organizations not operated for profit.
- Consumer AI accounts may allow model training on what staff enter; business plans usually do not by default.
- The safest way to use AI for donor communications is to draft with placeholders and merge real names and gift amounts in your CRM, so donor data never reaches the AI tool.
- The FCC has ruled that AI-generated voices are "artificial" under the TCPA, so AI voice calls follow prerecorded call rules.
Development staff are already using AI for donor communications: thank-you letters, year-end appeals and stewardship updates. The drafting help is useful. The exposure comes from the donor data staff paste in to get it, because your privacy policy and donation pages already promise donors where that data goes.
What privacy promises has your nonprofit already made to donors?
Most nonprofits have already made specific promises in writing: the website privacy policy, the line on the donation page that says you never sell or share donor information, gift acknowledgment language and any donor privacy pledge you display. AI does not change those promises. It adds a new place donor data can end up, and that is where organizations break them without meaning to.
Donor identities are treated as sensitive in federal law, too. The IRS says a tax-exempt organization is generally not required to disclose the names or addresses of its contributors when it makes its annual return available for public inspection. That exclusion does not apply to private foundations or section 527 political organizations. Donors who give expecting privacy have reason to expect it.
California adds a security duty. Civil Code section 1798.81.5 requires businesses that own or license certain personal information about California residents to maintain reasonable security, and section 1798.80(a) defines a business to include organizations operating whether or not for profit. Ask counsel how this applies to the specific data you hold.
Where does AI put donor data at risk?
The risk shows up when donor details go into an AI tool whose terms let the provider keep, review or train on what staff enter. Consumer and business versions of the same tool often have different terms. OpenAI, for example, says it may use content from individual ChatGPT users to train its models unless they opt out, while by default it does not train on inputs or outputs from its business products. Microsoft says prompts in Copilot covered by enterprise data protection are not used to train foundation models.
Everyday fundraising work creates the exposure. Common examples:
- A development officer pastes a major donor's giving history into a personal chatbot account to draft a stewardship letter.
- Someone uploads a full donor export to an AI tool to segment a year-end appeal.
- An AI meeting notetaker joins a donor call and stores the transcript on a vendor's servers.
- A CRM vendor switches on new AI features that read donor records under terms nobody at the nonprofit has reviewed.
If your privacy policy says you never share donor information with third parties, each of these can put you at odds with your own words.
How do you use AI for donor communications without exposing donor data?
Use AI to write the words, and keep the donor data inside the systems you already control. The steps below let staff get the drafting help without sending donor records anywhere new.
- Read your privacy policy, donation pages and gift acknowledgment templates, and list every promise you make about donor data.
- Approve one AI tool on business terms, used through the organization's own accounts, and tell staff it is the only one for donor work.
- Draft with placeholders. Ask the AI for a thank-you letter that uses fields like first name, gift amount and program, then merge real values in your CRM or mail tool.
- Never upload full donor exports. For analysis, remove names and contact details first or work with totals.
- Check AI features inside your CRM, email platform and notetaker before turning them on. Read the vendor's AI terms on retention and training, and confirm they match your privacy policy.
- Keep sensitive categories out of prompts entirely: anonymous donors, health or beneficiary details, and any gift tied to a confidential matter.
- Review every AI-assisted message before it goes out. Check gift amounts, names and dates against the CRM, and remove any impact figures the AI produced that your program team cannot source.
- Back the rules with controls: block personal AI accounts on organization devices, add data loss prevention for donor exports and turn on alerts when someone signs up for a new AI app.
If you are not sure which AI apps staff already use, start with finding the AI apps your staff already use before you write the rules.
Our guide to approving free AI tools without giving away donor data covers step 2 in more detail.
Do you need to tell donors you use AI?
Not necessarily, but your privacy policy has to stay accurate. Whether to announce that AI helps draft your letters is a judgment call. If an AI vendor will process donor information on your behalf, check whether your policy's language about sharing with third parties still describes what you do, and update it before the new practice starts.
AI voice is a separate case. In February 2024 the Federal Communications Commission ruled that calls made with AI-generated voices are "artificial" under the Telephone Consumer Protection Act. An AI voice calling donors falls under the same rules as a prerecorded call, so talk to counsel before using one in a phone campaign.
What should you do before year-end appeals go out?
Year-end giving is when staff are most likely to reach for AI under deadline pressure, so set the rules now. A short list for this quarter:
- Pull your privacy policy and donation page language and confirm they match how donor data actually moves.
- Name the approved AI tool for donor communications and move staff off personal accounts.
- Write a one-paragraph prompt rule: placeholders only, no exports, no anonymous donor details.
- Review AI settings in your CRM, email platform and meeting tools.
- Have board or leadership sign off, and record the decision. Our article on a one-page nonprofit AI policy can help.
- Brief volunteers who write donor notes on the same rules.
- Purge donor records you no longer need, so there is less to expose. Our nonprofit document retention policy guide covers which donor data to keep.
How does s90 help?
s90 sets up AI guardrails and new-app alerts for nonprofits, so staff use approved AI tools through organization accounts, personal AI sign-ups are flagged and donor exports are watched. If you want a second set of eyes on how your donor data moves before year-end appeals, we are glad to help.
