# How donors are using AI to decide who to support and what that means for your content Donors increasingly ask AI to vet a cause before they give. Here's what AI tends to get wrong about nonprofits, why your standard messaging isn’t effective with AI, and what a nonprofit can actually do about it. search experience. - Authors: Dewey Labs - Published: 2026-10-02 - Category: Guide - Canonical: https://www.askdewey.com/insights/how-donors-are-using-ai-to-decide-who-to-support-and-what-that-means-for-your-content/ --- A donor who previously searched for a nonprofit on Google often asks an AI tool instead today. This is especially true for younger donors, who lean on digital tools to validate a cause before giving. Regardless of whether the results are wrong, incomplete, or generic, the AI's answer is often the donor's first impression. Nonprofits need to plan for this, since their website, annual report, or gala are no longer reliable first experiences. ## What AI tends to get wrong about donation-related questions Ask a general-purpose AI tool like Gemini, ChatGPT or Claude what percentage of a nonprofit's donations go to programs versus overhead, and there's a good chance the answer is out of date, cites the wrong fiscal year, or doesn't match the organization's actual IRS Form 990. Unfortunately, this is a predictable result based on how these tools work. When a question falls outside of what the model knows about an organization, it fills the gap with something plausible-sounding drawn from its general training, not from that nonprofit's current numbers. This matters most for financial and impact questions, where accuracy is the point of asking. Questions about overhead ratio, program allocation, or how a restricted gift gets used, can generate results where plausible and even referenceable answers can be ten or twenty percentage points apart. ## The messenger matters as much as the message What surprises a lot of communications teams is how the same messaging can build or destroy trust based on whether it’s coming from a machine or a person. A December 2025 study in the [Journal of Theoretical and Applied Electronic Commerce Research](https://www.mdpi.com/0718-1876/20/4/341), led by researchers at the University of International Business and Economics in Beijing, tested this directly. Across four studies, the researchers found that when an AI chatbot delivered an abstract, values-driven message, donors found it less credible than the identical message delivered by a human, even when it used the kind of language a passionate human fundraiser might. In the case where they interacted with the chatbot, their willingness to donate dropped. The same values driven message from a human carried no such penalty. Interestingly, when the AI chatbot delivered a concrete, factual, procedural message instead, credibility and donation intent both went up. The explanation the researchers point to is a mismatch in expectations. People expect machines to be precise and factual, not moral or persuasive. When an AI tool tries to sound like a passionate advocate, it's playing a role donors don't believe, and trust drops the moment the mismatch becomes noticeable. When AI is answering questions on your behalf, concrete facts do more work than persuasive framing. The instinct to use messaging that you know works in-person, like making the appeal warmer or more emotionally resonant, can backfire. ## Citations earn donor trust Nonprofits need to make sure whatever AI says about their organization is both accurate and verifiable. An answer that states a program allocation percentage and links to the source document helps build donor trust. Near-universal citation lets a skeptical donor check the answer themselves. This also reframes what donor trust content should look like today. In addition to emotional appeal, nonprofits should ensure they have current, specific, sourced information for AI tools to draw from. A few places to start: - **Keep financial and impact facts current in a structured, easy-to-locate place.** A Form 990 filed once a year and buried in a PDF isn't doing much to prevent an AI tool from filling the gap with an outdated or generic answer. - **State the numbers donors actually ask about, directly and specifically.** Program allocation percentage, how a restricted gift is used, and what a given dollar amount funds should be clearly reported. If the specifics aren’t there, an AI tool will guess. - **Make the sourcing visible.** All facts about the organization's finances or impact should be cited, making it easy to see where that fact came from. - **Resist the urge to write donor-facing AI content as a persuasive pitch. **Language that is concrete and factual outperforms warm and aspirational when the messenger is AI. Dewey turns a nonprofit’s own current, verifiable content into a resource for donors. It delivers instant, accurate answers to your current and future donors’ questions, and never guesses based on outside information. Making this experience available on nonprofit websites builds credibility with current and future donors and ensures an accurate, branded experience. None of this replaces the human relationships that still drive major gifts. But for the moment when a donor is quietly checking an organization out before they've talked to anyone on staff, the AI’s answer is often the only version of the nonprofit they ever see.