(Illustration by iStock/Pikusisi-Studio)
Picture someone troubled by an article about growing numbers of children dying from preventable causes. Motivated to help, she decides to donate, but she is a charity novice and doesn’t know where to turn.
A few years ago, she might have opened a search engine and typed, “Where should I donate to save children’s lives?” The query would have generated a long list of links. The donor might have spent an hour clicking through charity ratings, reading annual reports, comparing local and global options, and weighing her values against the evidence. The path was imperfect, but it left room for discovery.
Now imagine the same scenario today. Bypassing search, our donor consults AI chat interfaces. One recommends donating to GiveWell’s top charities—Against Malaria Foundation, Helen Keller International, Malaria Consortium. Another recommends donating to GiveWell’s Top Charities Fund directly. The answers differ slightly, but the tools converge on the same destination. They do not surface other ways of having impact: community health workers supporting underserved families, early research on a pediatric disease, or a smaller global health group that the major charity evaluators have not yet rated. Both close with a version of the same line: “If I were making one donation, I’d make it this one.”
To be clear, this is a worthwhile answer: GiveWell does extraordinary work directing dollars to highly effective charities that save lives. The problem is that AI can redirect a broad human impulse to help to a narrow, confident answer before the donor has had a chance to discover what kind of giver she wants to be. No hedging. No alternatives. No reference to the thousands of other local and global organizations with track records of improving children’s lives.
Search used to hand donors a menu. AI hands them a verdict. The judgment is polished and confident, but it leaves out different nonprofits from one tool to the next. Well-funded local and national groups with steady press attention, third-party rankings, and tidy, structured data show up immediately in AI queries. Smaller or newer community-led organizations doing great work might be somewhere in the results, but they often stay hidden unless the donor knows what to ask for.
For the donor, what gets lost is the opportunity for individualized discovery—the chance to encounter novel approaches, stumble upon local actors, and weigh contradictory signals before settling on a personally meaningful choice.
This is a sea change for nonprofits that could become the next digital divide—not between organizations with websites and those without, but between organizations that AI tools recommend confidently and organizations whose credibility is still growing or lives mostly offline. Those in the sector who care about donors finding causes to support need to address this new visibility gap.
The Shift Is Already Underway
This concern isn’t hypothetical. Striking changes are already clear in at least three strands of data.
The first strand involves traffic to nonprofits’ virtual front door. M+R’s 2026 Benchmarks Study of 180 nonprofits found that organic search traffic declined consistently across 2025, a trend that the study attributes to AI-generated overviews that discouraged users from clicking through to sources. A similar study by Blue State found an even sharper slide, with organic search traffic to 24 nonprofit websites falling 35 percent in the first 10 months of 2025 compared with the same period a year earlier. People haven’t stopped searching for ways to give; they’re just ending up on fewer nonprofits’ sites.
The second strand is what the AI layer is doing in place of organic search. An evaluation by Whole Whale of multiple AI models recommending charities found a “herding effect,” with recommendation lists compressing toward small groups of big organizations that have been validated by third parties. Across cause areas, recommendations bunch around roughly 100 nonprofits within individual cause areas, even though the US social sector has more than 1.9 million registered organizations.
The third strand concerns how AI tools are influencing donors’ actions. Giving Compass has been running an AI-driven search experience inside Charity Navigator’s website that processed more than 22,000 queries during an initial three-month pilot. About 80 percent of those were cause-based rather than organization-based searches—which means donors were arriving with a problem to solve, not with a nonprofit already in mind. Within that experience, donors who used the AI search tool completed a donation at twice the rate of donors who used the site without AI.
The implication is important: When AI is well designed and embedded in the right place, it can accelerate giving decisions. But this raises the stakes of what AI tools recommend. A motivated donor scrolling quickly through a narrow list isn’t exploring—they’re converting. And if that list consistently runs to the same 100 organizations per cause area, the other nearly 1.9 million never enter the picture.
Take those three strands together and the picture is clear. Donor intent is still there, and they want a clear way to give. What’s changing is how that intention becomes a choice. The route from curiosity to donation is being redirected. Discovery used to be something the donor drove, clicking around, comparing, circling back. Now it’s mediated by an AI screen that decides what to surface and what to bury. The funnel tightens, options shrink; and once the tool feels accurate, donors tend to move quickly and accept what’s presented as the right answer. That’s a powerful dynamic and a serious responsibility. Who shows up in those results matters more than ever.
What Nonprofits Can Do
Nonprofits can’t solve these challenges on their own. But here are five ways they can address the AI visibility gap that do not require huge investments:
1. Begin with the basics. Keep the core facts about your organization easy to find, up-to-date, and presented in consistent ways. AI models cross-check. The description of your mission, programs, and outcomes on your website should line up with what appears on your Candid profile, your Charity Navigator listing, your annual report, and your tax filings. When AI models encounter conflicting language or data, they are likely to default to larger, well-documented organizations.
2. Don’t let third-party profiles drift. A few years ago, seals of approval from Charity Navigator or Candid felt like charming icons to park in the footer of a website. In today’s AI-filtered world, those profiles function more like the master record of a nonprofit’s credibility. When a donor asks an AI model who you are and what you do, that’s where it goes. An outdated profile on a third-party site isn’t a minor nuisance anymore.
3. Match your messages wherever you show up. Think press releases, partner websites, foundation databases, board biographies, and conference agendas. Every mention of your organization is a signal. Mixed messages register as static. Repetition and alignment read as credibility.
4. Run reality checks. Prompt ChatGPT, Claude, Gemini, and other AI models to sum up what you do, who you serve, and why a donor should support you. Look closely at what comes back. If the conclusions are incorrect, vague, or missing important pieces, see where your signals are weak. Some models offer ways to submit corrections on the spot—if so, take them up on it. Otherwise, correct the source of the flawed data.
5. Listen to the AI developers. Earlier this year, Google released a developer guide aimed at content for generative AI features. The guidance debunked many techniques popular in nonprofit marketing circles, such as keyword stuffing, special AI markup files, and slicing pages into AI-friendly chunks. Instead, it reinforced the importance of fundamentals: clear writing, correct details, and a steady, consistent presence in resources that carry weight.
What the Broader Ecosystem Can Do
If AI-mediated discovery becomes yet another burden for individual nonprofits to address, those most at risk—which tend to be those with the smallest budgets—will fall further behind. A sector-wide response is needed. Here are three things the sector can do:
1. AI labs should treat nonprofit discovery as a public-interest use case. AI developers should devote attention to how nonprofit information is sourced, attributed, and corrected. At a minimum, platforms should publish practical guidance on how organizations can appear prominently and accurately in query results. They could support open standards like the Model Context Protocol and other open data access points that make it easier for trusted sector data providers to expose high-quality nonprofit information through standard interfaces. Finally, labs should diversify their nonprofit data partnerships and include intermediaries that present local, community-led, non-English, and issue-specific organizations.
2. Funders should help nonprofits address the AI discovery challenge. AI visibility adds a new layer to the overhead problem: Nonprofits need staff and operational funding to keep websites current, maintain public profiles, and publish impact data. Approaching the problem one grant at a time will not work. Funders should underwrite shared resources: common guidance, AI discovery audits, and field-level backbones that help organizations stay visible and accurately represented.
3. Intermediaries should make nonprofit information more portable and more complete. Existing sector data platforms can move beyond profiles built mainly for human search and collaborate on portable data records that can travel across AI tools. Data providers should also get more creative about sourcing. They could draw from not only IRS records and contributed profiles but also local news, referral databases, public benefit directories, partner attestations, and emerging platforms that surface smaller or newer organizations.
As more people turn to AI to decide how to respond to social problems, the technology has a chance to share better information, surface innovative problem-solvers, and spur more giving. But AI tools will realize this upside only if they help people find nonprofits that are effective, trusted, and close to the people they serve—not simply those that are easy to document. Otherwise, AI may widen the distance between organizations whose credibility is already well-established online and those who are early in their journey or whose trust dwells in community.
Read more stories by Andrew Dunckelman & Dale Nirvani Pfeifer.