Your Development Office Should Lead the AI Policy Conversation

Register for the AI Policy Development webinar. 

AI is already inside your fundraising work, whether your organization has admitted it or not. You can see it in the draft that comes back a little too fast, a little too polished, and somehow less specific to the moment. And when donor communications start to feel like “everyone and no one,” the problem isn’t style. It’s trust.

We start with a scene from the HBO show Hacks where a tech investor tries to sell a comedian on using AI to generate punchlines. The counterargument lands hard: shortcuts can create something that looks like the outcome without being the work itself. From there, we bring that idea straight into nonprofit fundraising and ask the uncomfortable question: who should lead an AI policy and governance framework inside a trust-based organization?

We break down a real example of what happens when AI use collides with grief and institutional credibility, then zoom out to the quiet reality that most teams are already adopting AI tools ahead of any formal policy. We make the case that development leaders are the most qualified people to convene IT, legal, HR, and leadership not because we’re technical experts, but because we understand relationship dynamics, donor expectations, and what it costs when trust breaks. We also talk about why donor silence on AI isn’t reassurance, it’s assumed guardrails, and why that assumption raises the stakes for getting this right now.

If you want an AI policy that actually works, don’t settle for a template you can file. Build principles your team can explain, implement, and adapt as tools change. Subscribe, share this with a development leader, and leave a review so more fundraisers can step into the conversation.

Sign up for the June 24, 2026 webinar on leading your organization in developing an AI policy that works for your development office. 

👉 www.LetsTalkFundraising.com/AIPolicy

💡 Want to take the next small step?

→ Free Download: 12 Fundraising Prompts You'll Actually Use

→ Course:  The Fundraiser's AI Starter Suite

Transcript

Hacks and Shortcuts

Keith Greer 0:00

Are you watching that HBO show Hacks? If you aren't watching it, I'm not gonna spend a lot of time on this, but fix it. Because Gene Smart is extraordinary. The writing is sharp, and the series finale just dropped a couple of days ago, and I have basically been obsessing over it this last season. It's so good. But even if you've never watched a single episode, stay with me here because this is going to connect. And don't worry, I'm not going to give away any spoilers if you aren't caught up. This episode was from a few weeks ago, anyways, but still no spoilers, even if you're still at the beginning of season five. The show follows Deborah Vance. She's a legendary Las Vegas comedian. 50 years into a career, she has spent figuring out exactly who she is and how she works, and her co-writer is Ava, a millennial writer who came up through the digital age. And in this scene, they're sitting across from a guy, Silicon Valley investor, a real tech bro, named Graham, who wants to use them to train his AI platform to be a better comedy writer. His pitch is actually pretty sensible on the surface. He wants to start with wedding speeches, best man toasts, bridesmaid speeches, you know, the ones that are supposed to be funny and almost never are. He thinks AI can do it better. And Ava is immediately, viscerally opposed. AI is stealing creativity, draining the life out of things that matter. We're seeing this play out in real life with all the graduation clips of college grads booing anybody, talking about AI at commencement. Deborah, to Ava's complete surprise, is ready to sign the deal. And for a minute, you think, okay, this is Deborah being pragmatic, transactional when she needs to be, but then Graham tells her what he actually wants from her. Eventually, he imagines her using AI to write her own material, and Deborah pivots on her stilettos. He pushes on it. If you were stuck on a punchline and had a tool right there that could give you one, wouldn't you use it? And she thinks about it and she concedes, yeah, there's probably a shortcut there. But then she says, and this is the part that I keep coming back to. She says that if you use the shortcut, it's not the work. It's something else entirely. And her real argument, the one underneath it, is that becoming a comedian means you have to do it and fail. Do it and fail over and over until you figure out who you are and how you work and what actually is funny coming from you specifically. They AI can write a punchline, probably a pretty decent one, but it cannot make her a comedian. That only happens through the process itself. And you cannot extract the outcome from the process. Graham's final move is to leverage I can get anyone to do this. Kill Tony said he'd do it for free. And Deborah says calmly, yeah, you do that. And I've been thinking about that scene ever since that episode dropped a couple weeks ago because I think Deborah Vance just made the argument that development offices need to be making inside of their own organizations right

The Real AI Policy Problem

Keith Greer 3:16

now. So let's talk fundraising. Today I want to talk about AI policy, and I want to make a specific case that you, the development leader, are the most qualified person in your organization to lead this conversation about building one. Not because you're technical. If you're anything like me, you're probably not, but because of a set of skills that you've spent your entire career building, skills this conversation actually requires, that IT and legal and HR simply don't have. And I want to talk about what's at stake when development leaders step back from this and let someone else take it. I want to start with a real quick story and pay close attention to it because it's already playing out in trust-based

The Vanderbilt ChatGPT Email Backlash

Keith Greer 4:06

organizations. This one, a university, in exactly the category of communication that development offices navigate every single day. I'm going to do my best, Sophia Patrol here. Picture it, February 2023. Michigan State University has just experienced a mass shooting. Campuses everywhere are trying to figure out how to respond to their communities. How do you show up for students and staff who are scared and grieving from a distance? Vanderbilt University's Peabody College sends an email to students and staff, and it's a community care message about inclusion and support and the values of the college. It's signed by two administrators, Nicole Joseph, who led the Office of Equity, Diversity and Inclusion, and Hassina Moyedane, and I think I probably got that name mispronounced, but I hope it's close enough. She's assistant dean. And at the bottom of the email, there's a disclosure. It says, the message was paraphrased from OpenAI's Chat GPT. Students were outraged. And I want to sit with why, because the message itself was probably fine. The language was probably fine. But in a moment of real grief, in a moment that was asking the institution to show up as human, that disclosure made clear that what they received wasn't comfort. It was the simulation of comfort generated by a large language model, disclosed at the bottom almost as an afterthought. Nicole Joseph apologized. She called it poor judgment. She said that using ChatGPT for community communications in a time of sorrow and in response to a tragedy contradicted Peabody College's values. Both administrators temporarily stepped back from their roles while the incident was reviewed. Nobody in that office woke up that morning planning to cause harm. I genuinely believe that. They were trying to respond quickly to something impossibly hard, and they reached for a tool that was available, and they didn't have a framework for when that tool was and wasn't the right choice for a given kind of communication. The consequence was real. It was career-affecting, and it happened at a university in a moment of institutional trust, handled without the values question ever really being asked.

Your Team Is Already Using AI

Keith Greer 6:36

So let me ask you something. Think about the drafts that have crossed your desk in the last 30 days. The ones that came back a little faster than you expected, a little more polished than the person who sent them usually is, a little more generic than the situation called for. The language that sounds like everyone and no one at the same time. Your team is probably already using AI. Most teams are. The research on informal AI adoption in professional settings is pretty consistent. It's running well ahead of formal policy in almost every sector. People are using these tools because they're useful and because there's nothing in place telling them otherwise. And because the incentive structure rewards getting things done faster. So the question is not whether it's happening. The question is whether it's happening with any guardrails in place. And the question underneath that, the one this episode is really about, is who is responsible for making sure those guardrails are there? Let me

Why Development Owns Donor Trust

Keith Greer 7:43

start with something that I think most of us can agree on. Fundraisers are the guardians of the organization's financial relationships with donors, with foundations, and institutional funders, with the people who have chosen to place their trust and their resources in your mission. When trust breaks down in how the organization communicates with those people, development is accountable for repairing it. Not IT, not the communications team, though they'll be involved, development, because those are your relationships. And so development has more at stake in how the AI policy gets built than almost anyone else in the organization. Think about what a policy built without your voice actually looks like in practice. IT is optimizing for system security and technical compliance. And those are legitimate things to optimize for from an IT perspective. But a policy built around those priorities is not built around the dynamics of a major gift relationship. It's not thinking about the difference between an automated acknowledgement letter and a personalized note from a gift officer to a donor who just made the most significant gift of their life. It's not asking what it means for how we handle the notes an officer puts in the CRM after a vulnerable conversation with a donor about their estate. Those are your questions. They're not going to be asked by anyone else at the table unless you're there. And here's the part I want you to really sit with. A policy built without your voice doesn't just leave a gap. It can actively make things worse. If the policy is too restrictive and doesn't account for how fundraising actually works, your team will root around it. Not because they're careless, but because they're trying to do their jobs. And shadow AI tools being used informally, often with donor data being uploaded into systems that haven't been vetted by anyone, carry significantly more risk than formal, trained use would. It's harder to discover, harder to address when you find it. And when something surfaces, and in this environment it will, the breach isn't just operational, it's institutional. And development is on the hook for repairing the trust. Who builds the policy determines whether it actually works.

Leading Without Being Technical

Keith Greer 10:11

Now I want to push back on the assumption that I think is keeping most development leaders out of this conversation. The assumption is that leading an AI policy conversation requires technical knowledge. And so fundraisers who are not particularly technical by training or by identity, we disqualify ourselves. We say, this is an IT issue, they should own it, or I don't know enough to be in this conversation, or the more honest version, which is, I don't want to be in this conversation because I'll be out of my depth. And I understand that instinct, and I think it's wrong. The competency that this conversation actually requires is not technical knowledge. It's the ability to bring people from different backgrounds, disciplines, and competing priorities around a shared purpose. And you have been doing that your entire career. Let me tell you something from my own experience. I spent years raising money in the performing arts. And when I was talking with donors, what I realized pretty quickly was that I didn't need to know everything about the performing arts. I didn't need to know how to produce the show. I didn't need to understand the technical requirements for the lighting design, or have an opinion on how the director was thinking about the blocking. Those were experts doing work I wasn't qualified to do. My job was to understand who would love what this organization was bringing to our community and to help those people care about it enough to invest in making it accessible to everyone. Not to create it, to connect people to it. The same principle is at work here. You don't need to understand the technical architecture of AI platforms. You don't need to have an opinion on which tools are more or less secure at the infrastructure level. That's IT's job, and they should do that. You don't need to know the compliance requirements in granular detail. That's legal's job. What this conversation requires is someone who can convene a room full of people with different expertise and competing priorities around something that actually works. And that is a skill you have. Think about the coalitions you build as a matter of course. The president who has a different theory of what major giving looks like at your organization. The board member who thinks that every donor wants naming rights. The program officer whose definition of impact doesn't map cleanly onto how you're talking to donors about the work. You have been building bridges across those frames your entire career. You've been translating across those different worlds for years. Now you're pointing those same skills inward. Someone needs to convene IT and legal and HR and leadership around a policy that actually works for the people doing the relationship work. IT can't convene that conversation effectively. Legal can't. The CEO can mandate it by authority, but authority gets people into a room. It doesn't get them to a shared framework. The person who knows how to build a coalition around a shared purpose is you. We're not pointing the skill somewhere new. We're just pointing it inward. And your job in that room is not to become the AI expert in the building. It's the same job that you've always had. Make sure that whatever gets built reflects the experience of the person on the receiving end. In this case, it's the donor. Make sure that when IT is thinking about security architecture and legal is thinking about liability, someone is asking what this means for a gift officer trying to write a follow-up after a deeply personal conversation with a donor about their late spouse. That question is yours. No one else in that room is positioned to ask it. And I want to make one more argument.

Donor Silence Means Assumed Guardrails

Keith Greer 14:20

And I think that this is the one that changes the timeline. Because in 2022, maybe into 2023, donors were asking out loud whether organizations had thought through how AI would be used with their data, with their relationship information, with the communications that they received. And you could feel it at conferences. It showed up in the questions after panels. And some of you had donors raise it directly in conversations. Now in 2026, those same donors have largely stopped asking. And I want to be careful here because that silence could very easily feel like the concern went away. It didn't. What happened is that the conversation moved from the emerging concern category to the assumed to have been handled category. Donors aren't asking anymore because they assume that your organization has thought about this. They assume there are guardrails. They assume someone responsible is paying attention. But that silence is not reassurance. It is accumulated trust that most likely has not been earned yet. And when something goes wrong now, not if, because we are still in early organizations figuring this out and mistakes are going to happen, it is not just a mistake. It is a betrayal of assumed trust. The standard is no longer you made an error we can work through together. The standard is we trusted you to have handled this and you didn't. The longer an organization operates without a thoughtful framework, not a filed policy, but a framework that people actually understand and can apply, the higher the assumed standard becomes, and the greater the breach when something surfaces. I think about the Vanderbilt situation. That was 2023, three years ago, and the conversation we're having in 2026 is still about organizations figuring out what a responsible framework actually looks like. The gap between what donors assume is in place and what organizations have actually built, in many cases, that gap is big and it's significant. And your job is to close it. Not because somebody asks you to, but because you're the person in the room who understands what that gap costs when it becomes visible. So

Templates Create Rules Without Understanding

Keith Greer 16:47

here's what I want to offer you to you today, and also what I'm not going to offer you. I want to come back to Deborah Vance and that show Hacks for just a second, because at some point in the next few weeks, or maybe it's already happened, someone is probably going to hand you an AI policy template. Maybe it's already happened, a well-meaning colleague, an email from a professional association, something that floated through your LinkedIn feed, or maybe you researched it yourself. A document that covers the bases, says reasonable things about approved tools and data privacy and appropriate use categories. And the temptation, and I want to be honest about how real this is, is to file it, check the box, and move on to the 17 other things that need your attention this week. But Deborah understood something about the shortcut. It's not that the shortcut produces something terrible, it's that it produces something else. Something that looks like the outcome without being the thing. You get the object without the understanding that makes the object useful. If you copy paste a policy and file it today, I want to give you a challenge. Come back to it in 30 days. Sit down with a gift officer, your director of annual giving, someone on your team, and try to explain what's actually in it. Not what the sections are, why those guardrails exist, why this category of communication is handled differently from that one, why that particular line is drawn where it is. And no cheating, like you can't pull up the policy and look at it. Because if you can't explain it to your team and how they apply it, the moment the AI tools evolve, which they probably will before the end of this year, your team won't know how to adapt because they understand the rules, but not the principles. Rules are a snapshot. Principles are what scale. Rules tell your team what to do, and understanding tells them what to do when the rules run out.

Building Principles That Actually Scale

Keith Greer 18:47

That's the thing I want you to carry out of this episode because that distinction is exactly what the webinar I'm running on Wednesday, June 24th, 2026, at 2 p.m. Eastern, 11 a.m. Pacific is built around. I'm not going to hand you a policy. What I'm going to do is give you the framework and the roadmap to build one for your organization, with your stakeholders, in a way that you understand well enough to explain, implement, and adapt when the tools change and the situations your team faces don't map cleanly onto anything the policy originally anticipated. Because the process of building it, working through it with IT and legal, and your own team asking the hard questions together, deciding where the lines are and why, that process is not a step toward the policy. It is the policy. The understanding that comes from doing the work is the thing that makes it function. You cannot extract the outcome from the process. Sound familiar? The webinar is on Wednesday, June 24th, 2026, 2 p.m. Eastern, 11 Pacific, and you'll leave with a guidebook, something you will actually work through, not a document to file away, but a structured process for building your own framework for your own organization. So to register, go to letstalkfundraising.com forward slash AI policy, and I'll even put a link in the show notes for you. And if this episode resonated, I'd ask one more thing. Share it with a development leader who needs to hear it. The conversation that needs to happen in a lot of organizations right now is not happening yet. The more people who are ready to step into it, the better off we all are. And

The Donor Question You Must Prepare For

Keith Greer 20:36

I want to leave you with one picture before I let you go. Six months from now, someone on your team is in a conversation with a donor. Maybe the donor asks directly, How does your organization use AI in the communications you send to me? Maybe it's less direct. A moment of hesitation, a comment that signals the question is sitting just underneath the surface. And instead of your gift officer getting flustered or coming to you in a panic afterward, or quietly doing whatever felt easiest in the moment, they already know what to say. Not because they looked it up, because they understand why the guardrails exist. Because the principles make sense to them, not just the rules. IT has vetted the tools you're using, legal has reviewed how donor data is being handled, and your team knows the principles well enough to apply them to situations that nobody anticipated when the framework was first built. Because the tools are going to keep changing. New situations are going to keep arising that the policy didn't specifically address. And the only thing that scales across all of those changes is understanding. Your donors in that version of six months from now are on the receiving end of communications that reflect your organization's values. Its capabilities. Not just what AI made possible faster, but its values. That is what this work makes possible. And I think you are exactly the right person to build it. I'll see you next week, my friends. Bye-bye.