AI Is Already Here. Responsible Fundraisers Need to Pay Attention.

Image disclosure: This image was generated using artificial intelligence to illustrate a period in which telephone fundraising, the National Do Not Call Registry, flip phones and metaphorical Tyrannosaurus rex coexisted. It does not depict an actual fundraiser, telemarketer or prehistoric reptile engaged in charitable solicitation. Any resemblance to a Tyrannosaurus rex, living or dead, is entirely coincidental. No beneficiary likenesses or donor data were used in its creation. The limitations imposed by the subject’s unusually short arms are historically plausible but otherwise immaterial to the discussion.

I was Vice President of Client Services at a consulting firm specializing in direct mail and telephone fundraising when the National Do Not Call Registry went into effect. One of the key elements of that landmark legislation was that it generally did not apply to calls made by or on behalf of nonprofit organizations.

Keep in mind, we were still being charged by the minute for cell phone calls, the dinosaurs roamed the earth and the lava was still cooling.

But I remember the conversations clearly. From a compliance standpoint, the nonprofit exemption mattered. There was just one problem: try explaining it to someone who put their phone number on a Do Not Call list because they didn't want people calling them.

The law could say one thing while the donor's expectation said something else entirely. Fundraisers live in that gap more often than we acknowledge.

I thought about that recently when I stumbled across a new law in New York regulating the use of AI-generated people in advertising. And by “stumbled across,” I mean exactly that. I wasn't researching nonprofit AI regulation. I saw an article, thought Wait. What? and started looking for the fundraising conversation about it.

I couldn't find much of one.

That concerns me because while we're still debating whether nonprofits should or shouldn't adopt artificial intelligence, AI is already here. Our staff are using it, our vendors are building it into their products and our fundraising and communications teams can generate copy, images, video and increasingly realistic human beings with a few prompts. Now governments are legislating it.

Perhaps the most important thing nonprofits can do about AI right now isn't enthusiastically embrace it or categorically reject it. It's pay attention.

New York, AI and a person who doesn't exist

On June 9, 2026, a new New York law took effect requiring disclosure when certain advertisements use what the statute calls a “synthetic performer,” essentially a human performer created or altered using AI or software.

The law is primarily being discussed as consumer-protection legislation governing commercial advertising. Its operative language applies to someone engaged in dealing in property or services who, “for any commercial purpose,” creates an advertisement concerning that property or service.

Naturally, being a fundraiser, I had a different question: what happens when the synthetic performer is in a fundraising appeal?

The answer is: I don't know.

The statute doesn't explicitly say nonprofits are included or explicitly exempt them, and “commercial purpose” isn't defined in this section of New York law. There are good reasons traditional charitable fundraising might fall outside the provision, but there are enough unanswered questions that I would be uncomfortable simply saying, “Don't worry about it. That's an advertising law.”

Nonprofits have encountered this ambiguity before. Digital communications don't politely stop at state lines, state charitable-registration and solicitation requirements vary, and nonprofits don't always fit neatly into laws written around conventional commercial activity.

Now add rapidly evolving technology. Welcome to the fun.

But I'm less interested in whether New York could fine a nonprofit for using an AI-generated person than I am in what happens before we ever get to that question.

Why did you create that person?

Imagine a human-services organization wants to tell the story of someone receiving services but doesn't want to use an actual beneficiary's photograph. There may be excellent reasons not to: privacy, dignity, safety, confidentiality, trauma or consent.

So someone says, Let's generate an image with AI.

Okay. No actual beneficiary is exposed and we can create an image appropriate to the communication while protecting the people we serve. Except now I want to know: why did we create that person? Why that age, race or gender presentation? Why those clothes, that environment, that facial expression or that particular degree of visible distress?

Those aren't incidental choices. We are deciding how we want a donor to perceive the human experience we're representing.

Fundraisers already know that imagery and emotion affect response. “A sad picture raises more money than a happy picture” is an ethically complicated proposition when we're choosing photographs of actual human beings.

Generative AI changes the equation because we don't have to choose the photograph anymore. We can manufacture the person. 

More concerning, we can manufacture that person around the characteristics we believe are most likely to produce the emotional response associated with giving.

That's where, for me, the question moves beyond disclosure into something considerably more uncomfortable: when does persuasion become manipulation?

Persuasion isn't the problem

Fundraising is persuasive. We tell stories, select images, emphasize information, segment audiences and test creative choices. We intentionally create conditions under which generosity is more likely to become participation. That's not inherently unethical. It's fundraising.

But imagine discovering that a fictional seven-year-old girl looking directly into the camera with a particular expression generates more gifts than a fictional twelve-year-old boy looking away. AI increasingly gives us the ability to act on that information.

We could optimize the human being.

That sentence should make us uncomfortable.

And perhaps there's an earlier question: why does the communication require a human face at all? Does it genuinely help the donor understand the work and connect with another person's experience, or are we using it because a particular kind of face produces an emotional response that increases giving?

“It raises more money” cannot be the end of our ethical analysis.

Disclosure doesn't completely resolve the problem, either. We already use representative stock photography to protect beneficiary privacy, sometimes quite appropriately. But imagine an appeal telling Maria's story beside an intimate photograph of an exhausted-looking mother holding her child, except the woman isn't Maria. In fact, the woman doesn't exist.

We can disclose that the image is AI-generated, and I would argue transparency matters. But if knowing the person isn't real significantly changes the donor's response to the appeal, doesn't that tell us something about whether that information was material to the interaction in the first place?

Now we're talking about more than one New York statute. We're talking about trust.

The Wild West has Wi-Fi now

The synthetic-performer law interests me because it's one example of a much larger problem. The United States still doesn't have one comprehensive national privacy framework equivalent to Europe's GDPR. Instead, nonprofits navigate a patchwork of state privacy and charitable-solicitation laws, federal requirements, sector-specific regulations, vendor agreements, platform policies and evolving interpretations. Sometimes nonprofits are included, sometimes we're exempt and sometimes nobody seems entirely certain where we fit.

Meanwhile, fundraising technology has exploded: CRM systems, wealth screening, data append services, online giving platforms, email tracking, website pixels, predictive analytics, automated segmentation, meeting transcription, AI assistants, generated copy, images, voices and video.

Behind every shiny new capability is some combination of people, information and trust.

Which is why the debate over whether nonprofits should use AI is rapidly becoming beside the point. That ship has sailed, circumnavigated the globe and is sending us vacation pictures.

The more useful questions are what we're using these tools for, what information we're giving them, what happens to that information, what expectations donors and beneficiaries reasonably have and what obligations our choices create.

Data stewardship is donor stewardship

I've argued for years that donor data isn't simply an operational asset. Data is human. Every donor record represents a person and contains some portion of what that person has entrusted to us: identity, contact information, interests, relationships, giving history, behavior and an increasing amount of digital information gathered or inferred about them.

The technologies we use to collect, store, analyze and act on that information therefore aren't simply technology decisions. They're relationship decisions.

Privacy, security, data governance, technology governance and responsible AI use are all donor stewardship because stewardship isn't simply what happens after someone makes a gift. It's how we care for the relationship, the trust and the information people place in our hands throughout their participation with our organizations.

And that brings me back to paying attention.

We don't need to panic every time a legislature passes a technology law, ban AI or stop experimenting with tools that might make fundraising more effective, accessible or efficient. But we do need to pay attention to what those tools are doing, what information we're putting into them, what our vendors are doing with that information and what emerging laws may tell us about changing public expectations even when we're not certain they apply to us.

We also need to notice when efficiency becomes optimization and when optimization starts becoming manipulation.

The law will always lag behind technology. Compliance can tell us something about the minimum required of us, but it cannot tell us everything responsible fundraising requires.

I don't yet know whether New York's synthetic-performer law will ultimately be interpreted to apply to charitable fundraising. That's a legal question, and nonprofits using synthetic people in communications reaching New Yorkers should get appropriate legal advice rather than relying on a blog post from a fundraiser who remembers when cell phones had peak and off-peak minutes.

But I don't need a court to answer the question that interests me most: What information and technologies are we using to facilitate generosity, what obligations does their use create and are we exercising the professional judgment necessary to deserve the trust people place in us?

That's not really an AI question.

It's a fundraising question.