Should Your Nonprofit Use AI for Grant Writing?
Is using AI on a grant application cheating? Or is it just smart? If you've typed that question into a search bar at 11pm before a deadline, you're not alone — and the answer isn't as simple as yes or no.
I've wrestled with a lot of ethical dilemmas in my 20 years of nonprofit work. As an Executive Director, I walked the tricky tightrope between Board and staff. I dealt with employee situations that came with no roadmap. I turned down funding that came with too many restrictions — even in the middle of a budget crisis. And I navigated the Affordable Care Act, COVID-19, and plenty of smaller shifts that rocked our nonprofit and our community.
This is typical, because change is the purpose of nonprofits. If you sign up for a career in nonprofits, you're committing yourself to a constant tap dance on ground that's always shifting. You cannot stop the shift; you can only seek to become a better dancer. But sometimes the shift is so big that even great dancers stumble.
AI is this kind of shift. No one asked most of us whether we wanted AI, but we all have to decide now whether — and how — to use it. While plenty of people are championing, celebrating (and trying to sell you) AI, nonprofit leaders are quietly wrestling with its ethics. That's a serious concern, and an anxiety-churning one. One question I get asked often: should a nonprofit use AI for grant writing?
To me, AI is just a tool like any other. I can use a hammer to build a house that shelters my family. Or I can use a hammer to beat someone. The hammer itself isn't good or bad — it all depends on how I use it. In the past several years of my own wrestling, I've found that using AI for nonprofit grant writing really helps in some places and really hurts in others. Here are some considerations as you do your own ethical AI wrestling.
Where AI Genuinely Helps with Nonprofit Grant Writing
Fast research
AI is tops at fast research. I'm glad to kiss goodbye my days of spending two hours on 60 different Google searches to find the data points I need for a grant proposal (which, of course, was an improvement on the two days my mother spent in the library pre-Google). If I want to demonstrate the need for broadband access in a rural area, I ask AI something like, "Pull research-based or government-based data sources on the lack of broadband access in rural areas in Arkansas. Include links to all sources." Or, "Suggest research-based or government-based data on how lack of broadband access affects people in a community." Then I follow the links AI provides right to the sources. It cuts out so much time digging through the internet for the right information.
Funder research
AI is also a big help when I'm sizing up a potential funder. Before I write a single word of a proposal, I want to know what a foundation cares about, what kinds of organizations they've funded before, and how big their grants typically are. Now I can point AI to a funder's website and annual report and ask, "Summarize this foundation's giving priorities, typical grant sizes, and the types of organizations they've funded in the past three years." In a few minutes, I have a clear picture of whether we're a good fit, or whether I'm about to pour a week into a proposal that was never going to land. (I always confirm the details against the funder's own guidelines and their most recent 990 before going any further. More on that in a minute.)
Fast data analysis
AI is also great at fast data analysis. In the old days, if I downloaded a US Census spreadsheet with 2,000 rows and a bajillion columns and needed data for specific counties, it took a long time to wade through and pull out what was relevant. Now I can upload that same spreadsheet to AI and say, "Please pull out from this data set all the Arkansas counties located in the northeast part of the state whose poverty rates exceed the US national average." AI does it in about 30 seconds. I still have to double-check the data (AI does make mistakes), but the process is still 10 times faster than doing it manually. Even better, I can then say, "Make a table to insert into my grant proposal that shows all these counties with their poverty rates, compared to the US average." Goodbye, 15 minutes of formatting tables!
Charts, maps, and graphics
AI is great at making basic graphs, charts, and graphics for grant proposals. A map of your nonprofit's service area can be really illuminating to grant reviewers, but it's not a big enough job to justify sending to a graphic designer. I can ask AI to create a county-level map of Arkansas with X, Y, and Z counties highlighted. If I want to present my data as a bar graph, I can tell AI exactly how I want it to look. And if I'm not sure of the best way to present the data, I can ask AI for suggestions on the most compelling option. All of this strengthens a grant proposal's need section — without the hours grant pros used to spend fighting with Excel's complicated graph tools.
Making things look polished
Finally, AI is great at making things look pretty. I use it to format tables in my grant proposals (though I have to constantly tell it to set the font to at least 12 pt — apparently AI has great faith in my eyesight). I use it to create headers and page navigation elements in my Word doc after I've finished writing the copy. If you're not a pro at Microsoft Word, AI can even fix formatting issues faster than you can do it manually. My grant proposals look visually better than ever, thanks to AI.
One big caveat: check every number AI hands you
Here's the catch with everything above: AI is fast, but it isn't always right, and the data is where it most often gets things wrong. It will hand you a statistic with total confidence that turns out to be outdated, misread, or flat-out made up. It will give you a link to a source that doesn't say what AI claims it says, or that doesn't exist at all. It will do math that doesn't add up. So here's my rule: if a number or a source is going into a grant proposal, I click through and verify it myself. Every time. A reviewer who catches one bad stat will start wondering what else in your proposal is wrong.
Where AI Can Actually Hurt You
Here's where I draw a hard line: don't let AI do your writing for you. I know, I know — it's so very, excruciatingly tempting, especially if writing isn't your strong suit. You can write a sentence that sounds okay, and AI can take your plain sentence and make it sound so much fancier. Isn't that better? Won't it make you look smarter?
The answer is no. When you write a sentence, even if you're not a great writer, that sentence has your voice in it. The words you choose, the angle you are writing from, your lived experience, your passion for the mission — it all comes through to your funder. This gives the funder a clear idea of what makes your organization unique, even if there are technical errors in it. And when grant reviewers are reading a stack of 100 grant proposals, your job is to stand out. To be memorable. To be easy to read.
AI is designed to make everything sound like everything else. This means that if AI writes your grant proposal, it will sound exactly like the other 99 proposals in the stack and will likely use lots of fancy words that don't say much. The grant reviewer will roll their eyes and start skimming, because they've already read this proposal 99 times. And at the end, they still don't know why they should fund you over AI's other grant proposals.
Consider this paragraph, written by a real person to make the case for a program addressing homelessness:
Imagine you need $20 for gas to get to work. If you can't get to work, you'll lose your job. Many people would call a friend or family member for help.
But what if you didn't have any friends or family to call? You would lose your job. Because you lost your income, you can't renew your car insurance. You get pulled over and lose your driver's license. Then you lose your car. Without a car, you can't make it to a new job. With no income, you lose your home.
Where do you go when you find yourself at the bottom of this cascade of losses? On any given night in Benton, Carroll, Madison, and Washington Counties in Arkansas, 563 people are experiencing homelessness — 20% of whom are experiencing homelessness for the first time (Point-in-Time count, 2026).
Then I asked the AI to write a needs section paragraph for the same program. Here's what it returned:
Northwest Arkansas is one of the fastest-growing regions in the country, and its homeless population is growing with it. On the night of February 26, 2026, the annual point-in-time count identified 563 people experiencing homelessness across Benton, Carroll, Madison and Washington Counties, up from 507 the year before. That number has climbed steadily, from 436 in 2023 and 412 in 2024, and the 11% one-year increase continues a clear upward trend, with Washington County alone home to 63% of the region's homeless residents. Those affected are increasingly vulnerable: 146 people, or 26% of the 2026 count, were 55 or older, compared with 67 people (15%) in 2023. Among adults interviewed in 2025, 79% reported a disability and 17% were fleeing domestic violence. Rapid growth is driving housing costs up and pushing people out. Benton County has the fastest housing unit growth in Arkansas and Washington County the second-fastest, yet average regional rent reached $1,127.20 in the second half of 2025, 4.8 percent higher than the year before. Meanwhile, federal support has not kept pace: the Northwest Arkansas Continuum of Care receives about $700,000 a year, while local leaders estimate the region needs at least $2 million.
Editor's note: This AI-generated example is shown exactly as returned. Its figures have not been independently verified, which is exactly the point.
Be honest — did you read every word of this paragraph? (Your reviewers won't, either.) It's got a lot of relevant stats, but it doesn't really tell a story. It's tiring to read. It provides no context. It assumes the reader knows a lot that they don't (what's a point-in-time count?). And it's missing the critical ingredient of the first example: an understanding of the problem at a basic human level and an indication that our organization knows how to solve it.
And here's the kicker: look closely, and you'll catch AI contradicting itself. It says the number has "climbed steadily," then reports it dropped from 436 in 2023 to 412 in 2024. That's exactly the kind of error that slips past a tired grant writer at 11 pm and lands right in front of a reviewer. Verify everything.
This gets to the heart of the great AI pitfall: letting AI do your thinking for you.
AI does not know your program, your community, or your cause in the real-world way that you do. It cannot design your program, build relationships with your funders, or develop a grant strategy based in your reality. These are hard things that take time, and I understand the yearning for someone (something) else to step in and do it for you. But it does not work. Judgment is still yours, and cannot be replaced.
So, Is Using AI on a Grant Application Cheating?
So let's go back to the question we started with. Is using AI on a grant application cheating? My honest answer: using AI as a tool isn't cheating. Letting AI do the work that's supposed to be yours (your thinking, your voice, your program) is where it starts to cross a line. And as we've just seen, it's also bad strategy.
That said, funders get the final say, and more of them are weighing in. A growing number of foundations now publish their own AI policies, and some federal agencies have, too. The National Institutes of Health, for example, has said it won't consider applications substantially developed by AI to be the applicant's original ideas. So before you start, read the funder's guidelines carefully and look for anything about AI. If they ask you to disclose your use of AI, disclose it. If they prohibit it, don't use it. And if you're not sure, ask your program officer. That's a relationship-building conversation, not a red flag.
One more caution: be careful what you feed AI. Don't upload confidential information into a public AI tool: client stories with identifying details, donor records, or staff and financial data you wouldn't post publicly. Protecting the people you serve is part of the job, too.
A Simple Rule to Keep AI in Check
Here's the rule I use:
Research, analysis, presentation, formatting: hand those off. Your voice, your story, your strategy, and what you choose not to say: keep those.
What AI Cannot Replace
AI can't replace the work of an experienced grant professional — someone who can help with program design, grant budgeting, funder cultivation, overall strategy, and weaving grants into the rest of your fundraising. And it can't replace a skilled writer who brings a real voice to the page, pays attention to readability, stays faithful to the community you serve, and keeps every proposal strategically aligned.
Not sure your organization is ready for its next grant? Start with our free Grant Readiness Checklist. It walks you through what funders look for before you ever open a blank page (or an AI chat window).
Want a Second Set of Eyes? (Human Ones.)
You don't have to figure this out alone, and you definitely don't have to figure it out at 11 pm before a deadline. If you'd like a second set of eyes on your grant strategy (human ones, I promise), let's talk. We'll look at where you are, what's realistic for your team, and where AI can actually save you time without costing you your voice.