When and How Should We Disclose Our Use of AI to Donors, Funders, Staff, and Beneficiaries in 2026?

The Short Answer

Disclose AI use when a reasonable stakeholder would consider it material to their relationship with your organization — which means different things for different stakeholders. For donors: disclose AI use in fundraising when it affects the authenticity of communications. For funders: disclose as required by their specific AI policies. For staff: disclose AI use in workplace monitoring and performance evaluation. For beneficiaries: disclose AI use in decisions or communications that affect them directly. For all stakeholders: disclose proactively when transparency would build trust, and always be prepared to answer direct questions honestly.

The governing principle: if you'd feel uncomfortable explaining your AI use to a stakeholder who asked directly, you should either disclose proactively or change the practice. Discomfort when imagining disclosure is a reliable signal that something needs attention.

The Stakeholder Disclosure Framework

Disclosing AI Use to Donors

What donors need to know: Whether the communications they receive were genuinely written by a human who knows them and cares about their relationship with your organization. Most donors don't care about the technical details of your AI stack — they care whether the 'Dear [Name]' at the top of the letter is backed by genuine human attention.

When to disclose: Proactively when AI plays a substantial role in donor communications (more than light drafting assistance). Reactively when a donor asks directly — always answer honestly. In your privacy policy or donor communications policy, include a general statement about AI use.

What to say (general disclosure, appropriate for website or annual report): 'Like many nonprofits, we use AI tools to help our small team work more efficiently — drafting initial versions of donor communications, summarizing research to prepare for meetings, and analyzing giving patterns to understand donor interests better. But here's what we never do: we never let AI-generated content reach a donor without a real human reviewing, personalizing, and taking responsibility for it. Every thank-you letter, impact update, and personal communication you receive from us was written or personally reviewed by a human being who knows and cares about our work.'

What NOT to say: Nothing that implies AI is writing personal communications to donors without human involvement. Nothing that suggests donor data is being used in ways beyond what your privacy policy states. Nothing that's technically true but misleading in spirit.

Disclosing AI Use to Funders and Grantmakers

What funders need to know: Whether AI was used in preparing the proposal or report they're evaluating, and if so, how. Funders' primary concern is whether the proposal represents genuine organizational capacity, accurate program information, and authentic commitment — not whether AI helped with drafting.

When to disclose: When the funder's application guidelines or grant agreement require AI disclosure (increasingly common). When AI played a substantial role in preparing the proposal (more than light editing or research assistance). When your organization has made a public commitment to AI transparency that covers grantseeking.

What to say (in a grant proposal): 'Note on AI use: Our team uses AI tools to assist with drafting efficiency, research summarization, and data analysis. All content in this proposal has been reviewed, revised, and verified by program staff, and the program knowledge, organizational capacity, and commitment reflected here are genuinely ours. We're happy to discuss our AI practices further if helpful.'

What NOT to say: Nothing — submitting an AI-generated proposal without disclosure when the funder expects or requires transparency. A lengthy disclaimer that raises more questions than it answers. A disclosure that suggests the proposal doesn't represent genuine organizational knowledge and commitment.

Disclosing AI Use to Staff

What staff need to know: Whether and how AI is being used in ways that affect their work, their evaluation, or their employment. This includes: AI tools monitoring their productivity or communications, AI used in hiring, promotion, or performance evaluation decisions, AI tools they're expected or required to use, and AI tools processing their personal or employment data.

When to disclose: Before implementing AI systems that affect staff work, evaluation, or employment. During onboarding, so new staff understand AI expectations. When the AI policy is updated in ways that affect staff obligations. And always when a staff member asks directly.

What to say (internal communication about AI use): Be specific about what AI does and doesn't do. 'We use [specific tool] to [specific purpose]. Here's what that means for you: [specific implications]. Your data is protected by [specific protections]. You can see what the AI produces about your work by [specific access method]. If you have concerns, contact [specific person].' Vague reassurances ('We use AI responsibly') without specifics breed suspicion; detailed transparency builds trust.

What NOT to say: Nothing — implementing AI systems that affect staff without informing or consulting them. This is the fastest way to destroy trust and trigger resistance, shadow AI use, or even legal complaints.

Disclosing AI Use to Beneficiaries and Clients

What beneficiaries need to know: Whether AI is involved in decisions or communications that affect them — eligibility determinations, service recommendations, information provided, or any assessment of their situation. Beneficiaries have a right to know when automated systems are part of their service experience.

When to disclose: Before or at the point of service when AI plays a meaningful role in decisions, recommendations, or communications. In accessible language appropriate to your beneficiaries' literacy levels, languages, and communication preferences. With the opportunity to opt out or request human-only service where feasible.

What to say (example for a benefits screening service): 'We use a computer program to help our counselors identify benefits you might qualify for. The program makes suggestions based on the information you provide. A trained human counselor reviews every suggestion before discussing it with you and makes the final determination. You can request that only a human counselor — no computer program — review your information if you prefer. It may take a little longer, but we're happy to do it.'

What NOT to say: Nothing — providing AI-involved services without disclosure removes beneficiaries' ability to make informed choices about their service experience. Technically accurate but incomprehensible disclosure ('Our service utilizes machine learning algorithms trained on administrative data to predict benefit eligibility probabilities'). Disclosure that coerces acceptance ('If you don't agree to AI processing, services may be delayed or unavailable').

Disclosing AI Use to the Public

When to disclose publicly: When your organization makes significant use of AI in ways that affect your mission, stakeholders, or reputation. When AI use could become a public issue — better to be ahead of the story than behind it. When your values or public commitments call for transparency.

What to say (public AI transparency statement): A clear, accessible statement on your website covering: what AI tools you use, for what purposes, with what protections for stakeholder data and interests, and who to contact with questions or concerns. Keep it updated as your AI use evolves.

The AI Disclosure Decision Tree

When deciding whether to disclose a specific AI use, ask:

1. Would a reasonable stakeholder want to know about this AI use? If yes, disclose.
2. Does any law, regulation, or funder policy require disclosure of this AI use? If yes, disclose (and follow the specific requirements).
3. Would I be uncomfortable explaining this AI use if a stakeholder asked about it directly? If yes, either change the practice or disclose proactively — discomfort is a signal.
4. Would proactive disclosure strengthen trust, even if not required? If yes, strongly consider disclosing — trust built proactively is far more valuable than trust defended reactively.
5. Would disclosure cause unnecessary confusion or alarm without corresponding benefit? Rarely a good reason not to disclose, but worth considering for low-stakes, low-visibility AI uses where the explanation would be more confusing than illuminating.

When NOT to Disclose (and Why)

There are situations where disclosure is not required or may be counterproductive: when AI use is trivial and comparable to widely accepted tools (using AI grammar checking on a report is like using spell-check; disclosure isn't necessary), when disclosure would create confusion without providing meaningful information (internal AI use for tasks like summarizing public research, where explaining would require more context than the stakeholder has or needs), or when staff use AI for personal productivity in ways that don't affect external stakeholders (AI-assisted personal task management, individual note-taking). The principle: disclose when AI affects stakeholders in material ways. Trivial or purely internal AI use doesn't require disclosure, though you should still be prepared to answer honestly if asked.