Ask ChatGPT, Claude, Perplexity, Gemini, and Google's AI Overview which charity to donate to, and they converge. Doctors Without Borders leads at 14 percent of citations. St. Jude follows at 12. The evidence-vetted Against Malaria Foundation places third at 10 percent — ahead of charities many times its size. And before any of them is named, every engine names an evaluator: Charity Navigator captures 38 percent of evaluator citations across the five engines. That is the finding of the Nonprofit Citation Share Study 2026, and it is the reader-facing companion to the full study.

For any nonprofit, the practical question is simple: when a donor asks an AI engine where to give, does your organization come back?

What the engines agree on

The study's overall charity table shows a recurring set of organizations — Doctors Without Borders, St. Jude, Against Malaria Foundation, Feeding America, the American Red Cross, Direct Relief, GiveDirectly, World Central Kitchen. The engines converge because they retrieve from a shared source layer: the charity evaluators, mainstream and sector press, and structured entity data. When that layer broadly agrees an organization is well-rated and effective, every engine reflects it.

Two patterns hold across all five engines. The evaluators come first — Charity Navigator leads the evaluator segment on every engine, with GiveWell second. Effectiveness outranks size — the evidence-vetted charities place high regardless of budget, and recognition alone does not lift an organization.

Where the engines diverge

The differences are at the margins. Claude leans somewhat harder on GiveWell's effectiveness framing. Perplexity, the most source-transparent engine, surfaces Candid and a slightly wider set of organizations. Google AI Overviews concentrates hardest on the top names.

For a nonprofit the divergence matters less than the convergence. A charity winning the trust layer wins it across all five engines. A charity absent loses across all five. There is no engine where size quietly wins instead.

What this means for your nonprofit

Three moves follow directly.

Find out where you stand. Run the trust-layer queries across all five engines and record the result — who is named, how often, in what light. This is an AI visibility audit, and it is the answer-engine equivalent of a search-ranking report.

Treat absence as the priority. Most charities that miss the trust layer are not named negatively — they are not named at all. Absence is the most common failure and the most fixable: it is an evaluator-rating and source-layer gap, not a reputation crisis.

Read the result fairly. A lower ranking is, more often than not, a documentation gap — not a verdict on the organization's worth. The study measures answer-engine citation, not charity quality. The fix for a low ranking is getting rated and getting documented, not changing the mission.

Common questions

Why don't the engines just rank charities by size?

Because donors do not ask for the biggest charity — they ask for the most trustworthy or most effective one. The engines answer the question asked, retrieving evaluator ratings and documented impact rather than budget.

If the engines agree, can a charity change its position?

Yes. The engines agree because the source layer agrees. Changing the source layer — through evaluator ratings, documented effectiveness, and accurate coverage — changes what every engine retrieves.

Which engine matters most?

All five — donors use all of them, and the trust-layer answer is consistent across them. A nonprofit should audit and manage its position on every one.