It's Been a Minute Since I Wrote About AI

The last time I wrote about artificial intelligence on this blog, it was 2023. 

Holy smokes! 

AI in our world has changed just a little bit since then. In 2023, generative AI still felt like a novelty act at the edge of the nonprofit field, something nonprofit folks whispered about at conferences with a mix of curiosity and dread. 

And secretly, no one really wanted to admit they used it.  

Fast forward to 2026, and AI adoption in the nonprofit sector has gone from a fringe experiment to nearly universal practice: a 2026 benchmark study of 346 nonprofits found that 92% now use AI tools in some capacity. Yet only 7% reported major improvements in their ability to achieve their mission — a stark reminder that widespread adoption does not automatically equal meaningful impact. That stat alone tells you how quickly the ground has moved beneath all of our feet. (Virtuous, The 2026 Nonprofit AI Adoption Report.)

I want to be honest about the journey it took me to get here, because I do not think I am alone in it. This is not a just-adopt-the-tools pep talk, nor is it a doom-and-gloom warning about the future of our work. It is certainly not an argument meant to persuade anyone that they should or should not use AI.

Instead, it is a reflection on the questions I believe we all need room to ask: How do we use AI? When is its use appropriate? When should we choose not to use it? What safeguards do we need to protect privacy, accuracy, relationships, and authentic voice? 

These questions do not exist in a vacuum. 

Bigger questions not discussed in this blog include legitimate concerns about AI’s environmental footprint, particularly the electricity, water, land, construction materials, and hardware required to build and operate data centers. And not all AI-related water use is equal. Some data centers use sealed, closed-loop, non-evaporative cooling systems that recirculate cooling fluid and greatly reduce (or nearly eliminate) ongoing water use for cooling. Others rely partly or primarily on evaporative cooling systems, which require continuing makeup water. The specific system design, local climate, and source of water all matter. 

These bigger questions also raise more human-centered questions about equity, bias, ownership, transparency, and job loss. When AI significantly changes or replaces someone’s work, we should be willing to pause and ask whether the outcome is truly better than the work performed by the person who held that role; not simply whether it is faster or less expensive. The answer likely depends on the sector, the task, who bears the risk of error, and whether AI is being used to support human expertise or to eliminate it. 

Spoiler alert: I do not have a final answer or a neatly personal resolved opinion on all of these questions. I am not going to pretend that I do. What I can do is stay curious, keep learning, make more conscientious choices about my own consumption when possible, and protect what matters most in my work: my authentic voice and the voices of my clients, especially in grant narratives.

For me, learning to live and work in this evolving landscape has meant moving beyond both fear and enthusiasm. This blog post is a reflection on how I moved from fear of getting left behind into real questions about ethics and authenticity in my writing, its overall effect on the nonprofit sector (the sector I serve the most), and eventually into a grounded, intentional set of choices about where AI fits into my practice and where it absolutely does not.

So let’s go…

The Fear Phase: Getting Left Behind

When ChatGPT and its cousins first exploded into public consciousness, my gut reaction was anxiety. As a grant writer whose entire value proposition rests on crafting authentic, relationship-centered narratives for funders, the idea that a machine could generate a "complete proposal" in seconds felt like a threat to the whole model. I mean the sheer number of LinkedIn connections stating they could “save me hundreds of hours in grant narration for $100 a month” was staggering in 2025, and even now, I still get one at least a week. 

I wasn’t alone in that fear. 

The Virtuous Report (full citation and link below) captured a workforce moving quickly without much of a roadmap. By late 2025, 81% of nonprofits reported using AI individually and on an ad hoc basis, without documented workflows or shared team practices. In other words, many of us were experimenting quietly, trying to understand the tools. Frankly, we were all just hoping not to fall behind, or worse, become irrelevant.

That fear pushed me to start testing tools, mostly in secret at first, the way you might sneak a peek at a competitor's website. What I found was uncomfortable and hard for me to grasp: the tools I did understand were genuinely useful for certain tasks, and ignoring them entirely wasn't a sustainable strategic position for a consultant serving nonprofits competing for the same shrinking pool of funding. 

It turns out that statistics from recent nonprofit research on AI usage back up what I was feeling: In 2025, 60% of nonprofit professionals were strongly interested in using AI to strengthen grant writing and fundraising, while 24.6% were already using it for grant writing specifically. (NonProfit PRO, How Artificial Intelligence Is Changing the Nonprofit Sector)

The fear of being left behind, it turned out, was shared by nearly everyone in my nonprofit network…we just weren’t saying it out loud.

But the grant consulting community was most definitely saying it, and pouring resources into helping us navigate and understand it. I mean, just a year ago at a conference (featured photo) where a colleague and I spoke about grant processes, a central question asked of us over and over by non-profits was how to answer “Did you use AI in this grant proposal?” from a funder. Which I have seen less and less of since January 2026, but still it has merit. I discuss this facet more later on in the blog. Regardless, cultural acceptance of AI’s use in 2025 and into 2026 was shifting FAST.

The Reckoning: Questioning Ethics and Authenticity

I have learned that fear is a poor long-term motivator. Once the anxiety settled (about 6-7 months), a much harder set of questions surfaced. What does it mean to use a tool trained on scraped internet data to write on behalf of organizations serving marginalized communities? What happens to the authentic voice of an executive director or a program participant if a machine is generating their story? And practically, will funders even accept work touched by AI?

That last question turned out to have some teeth.

Will Funders Even Accept This?

Candid’s 2024 survey captured a philanthropic sector that was still largely uncertain about AI-generated grant proposals: 23% of foundations said they would not accept them, 67% had not yet decided, and only 10% said they would. By 2026, AI itself is far less novel, and as nonprofit use has become widespread, funders are increasingly discussing data responsibility, outcomes, and long-term sustainability. Yet there is no newer, comparable data showing that foundations have broadly changed their formal acceptance policies. 

No real data (yet), so let’s follow the action because many private funders are closing applications early due to the rising number of applications, which tells you the grant world has shifted. Some of this may be funding channel shifts, but it is also most likely AI-related too. I mean somebody is taking these LinkedIn grant-generating peeps up on their offers? The most responsible conclusion I can make is that funders have not reached industry consensus, but the uncertainty identified in 2024 remains relevant: grant writers should follow each funder’s explicitly stated guidance, protect confidential information, and ensure every proposal remains accurate, human-reviewed, and rooted in the organization’s authentic voice.  

On the flip side, with grant review, some funders are starting to explore whether AI could help them manage larger volumes of applications they receive or speed up the review process. Candid’s 2025 foundation survey found that 97% of foundations were not using generative AI to screen applicants or make funding decisions, 1% said they were, and 2% were unsure. Looking ahead, 19% were considering using it, 3% expected to use it, and 12% were uncertain. 

This means more funders might look into using AI to combat the application influx that continues at such a high rate. 

Using It Without Using It Well

While funders sort out their own AI policies and decide how they want to respond to the large industry adoption of AI, I felt it was important for me to distinguish between using AI and using it with intention. And non-profits are starting to ask this too.

Let’s face it, when only 7% of nonprofits describe their use of AI as strategic enough to produce substantial improvements in their ability to achieve their mission, you kinda want to pause and ask “what are we doing here?” In fact, most organizations are still experiencing what the researchers call an “efficiency plateau”: faster drafts, quicker responses, and time saved on routine tasks, without a corresponding transformation in organizational capacity or outcomes. (Virtuous, The 2026 Nonprofit AI Adoption Report.)

So that hurts. 

I personally want more free time in my brain and body, and if you lead or work within a nonprofit, I know you want more time to create a bigger impact for the communities you serve.

Ugh, the last thing nonprofit leaders need is feelings of more isolation.

Intentions Aren’t Policy

Using AI intentionally is what frees up your time for the human things. But to get there, we've got to start writing down our intentions to create policies. Intention is individual, but for the organization as a whole, we look at governance. And governance too remains a serious concern. 

The same 2026 study found that 47% of nonprofits had no AI governance policy, while TechSoup and Tapp Network’s 2025 benchmark report found that only 24% of surveyed nonprofits had developed a formal AI strategy. Together, these findings suggest that many organizations are adopting tools faster than they are establishing the policies, accountability structures, and shared practices needed to use them responsibly. 

I am personally working through this with my team after hearing a fantastic presentation at the Learn Grant Writing Gathering Conference by a fellow grant consultant who lives in Germany, where they have very strict AI compliance rules. 

Wildly enough, even all that fear, data, and noise didn't scare me out of using AI. 

Instead, it clarified exactly where the lines needed to be drawn for my own practice.

Where I Say Yes

After working through the fear and the personal ethical reckoning, I landed on a set of AI uses that genuinely make my grant writing practice stronger without compromising the integrity of the work or the trust funders and my clients place in it. If this helps you, please use it, tweak it, or ask your own set of questions, especially if you lead a nonprofit.

My yeses align with broader sector data showing the top ways grant writers use AI, which directly speaks to drafting support, editing, and research rather than final creative output. For example, checking grammar, spelling, and punctuation (53%), brainstorming headlines (53%), and generating first drafts of content (39%) top the list of how nonprofit professionals actually use AI chatbots. (Julep CRM. “From Curiosity to Commitment: How Nonprofits Are Using AI in 2026 and What Has Changed in the Last Year.) 

Here's where I've chosen to say yes:

• Research for grant proposals: Using AI to research funder priorities, sector data, and comparable program models, with more speed and efficiency. I always cross-check against primary sources before any facts or statistics enter a narrative. For more complex grants, I still try to purchase as many journal-reviewed articles or open-resource verified information as possible, and then upload them into my project spaces to use.

• Closed systems only: I limit sensitive or proprietary work to platforms where I control the login, data handling, and guardrails, rather than open, general-purpose tools that may use the data to train their models. I require this of all my contractors; in fact, we recently updated our systems to be closed and internal. 

• Grammar and mechanical checks: A majority of nonprofit professionals already use AI for this as it is low-risk and high-value. However, AI does not replace good human editing; that is a whole other skill set highly valuable to grant work. Also, I still love my Grammarly too (which will trigger an AI usage, if you are concerned, but it is so much better than pulling out the Little Brown Handbook, yep, I am THAT age). 

• Catching missed logic model links: I love using AI as a second set of eyes to flag gaps between stated activities, outputs, and outcomes before a proposal goes out the door. Great to cross-reference with a rubric if you have a copy.

• RFP guideline checks against final applications: This is a compliance pass to confirm that the final draft actually answers every requirement in the funder's guidelines, reducing the risk of technical rejection.

• Idea generation for marketing copy: I enjoy using AI as a brainstorming partner for campaign angles and messaging concepts, but not as the final voice. It's more like overcoming the fear of a blank page. I also enjoy taking my 500-word caption and asking AI to condense it to 100 characters; I mean, it is insulting, but I like it. 

• Finding the best place to insert research into a narrative: AI is helping with identifying where a statistic or data point will land with the most persuasive weight, without letting it write the persuasion itself.

Where I Say No

This is just as important as where I use AI: where I deliberately and intentionally keep it out. My practice is built on data and logic, and the felt, human dimensions of a funding story, and there are places where a machine simply cannot and should not stand in for that.

My Noes come from my experience of what I see works well in good grant writing, and it is easy to spot when a story sounds fake, falls flat, or is unethically written. Also, more importantly, funders can tell the difference too. 

• Writing the actual narrative: The ethical storytelling core of a proposal stays entirely human-authored, because the people and communities behind these stories deserve to be represented by someone who has sat with them, not a language model. AI can refine, but it cannot replace. (I know a lot of people disagree, and maybe over time this will change, but I am going to stand my ground for now.)

• Replicating an organization's authentic voice: Each nonprofit has a unique way of describing its impact, programs, and population it serves. This unique voice is earned through relationship and institutional memory; outsourcing it to AI risks flattening exactly what makes a funder trust an applicant. (But AI is getting close, especially Claude, but even so, it isn’t enough.)

• Grant prospecting: Okay, this one is a little mixed, but hear me out. Grant prospecting has two parts: finding information and strategizing. AI can quickly generate possibilities and find relevant data quickly (like this funder gave to similar funders to your organization in 2023), but a long list of foundations is not the same as a thoughtful funding strategy. I employ a part-time researcher (an actual human person) focused on vetting new leads for true alignment, including eligibility, priorities, past giving, grant size, and real relationship potential, because that human discernment is essential to building a sustainable funding pipeline. (Instrumentl has a great article on this; I will pop in at the bottom.) 

• Creating graphic design: Visual identity and design communicate an organization's values in ways that require human creative judgment and cultural sensitivity that current AI image tools don't reliably deliver for this purpose.

  
  

Side note on design: Okay, I get that many small nonprofits are loving AI- generated graphic design, and maybe it will improve. I personally do not care for it, but I understand the cost savings. When your financial margins are tight, and it comes down to hard choices, we do what we have to do, but please, at least try Canva. Those AI-designed flyers are not doing your mission any favors. 

Where My Yeses and Noes Meet

My “yes versus no” line isn't arbitrary. It mirrors an emerging best-practice consensus in the sector: use AI to generate drafts and conduct research, but ensure every submission reflects genuine human review, accuracy, and a voice that authentically represents the organization's actual relationships and track record. 

Oh, and let me tell you, as someone who loves their words, I often enjoy seeing AI make everything longer, lol, right? I know, even I am laughing at myself right now cause I think this is the longest blog I have ever written.

BUT… have you ever noticed how, after a while, it all just sounds the same? That is because it does; longer isn't always better. In fact, sometimes it is just plain repetitive. 

The human review layer isn't optional; it's foundational.

Why I Still Choose to Use AI

I could have stopped at "no" to everything, and for a while, fear nearly took me there. But sector and non-sector data consistently show that organizations sitting out AI entirely aren't gaining an ethical high ground so much as falling behind operationally (there is a “but” coming). 

Business adoption of AI surged from 55% in 2023 to nearly 80% in 2025, and the nonprofit sector appears to be following a similar upward trajectory. (Northern California Grantmakers, Equitable AI; they were summarizing a broader finding from the Center for Effective Philanthropy’s national research.) 

And better-resourced nonprofits are already adopting AI at nearly twice the rate of smaller organizations, deepening a resource gap that disproportionately affects the exact grassroots and community-serving organizations our communities rely on. 

But…to really make a positive impact,  AI adoption needs to be intentional; otherwise you just spin your wheels without really going anywhere.

So, what really changed for me?

At the time of this blog, I am 48. I am at the tail end of Gen X. I remember life before cell phones, when the first computer entered our house, and when I had to sit in a training to learn how to use email (assuming I even decided I wanted an email address because it was OPTIONAL). I remember handwritten papers, trips to the library that lasted hours seeking research, and doing statistics by hand. And while those experiences taught me persistence, resourcefulness, and the value of doing the work, I am grateful that technology has made many parts of my work (and life) more accessible and efficient.

What changed for me was realizing that I do not have to choose between resisting every new technology and handing my work over to it. I can use AI within a framework grounded in my values: clear guardrails, closed systems for protected information, a human-authored voice, and limited support for research, mechanics, and compliance checks. 

But… 

There is a hard stop when it comes to meaning-making and telling someone else’s story.

And...

I accept that I have not resolved every ethical tension I have around AI. 

And...

I do not think I should. 

Some discomfort is worth holding onto, especially when deadlines are tight, and shortcuts feel tempting. Because in grant writing, and also in a lot of nonprofit work, faster is not automatically better; the work still has to be accurate, aligned, authentic, and worthy of a funder’s trust.

That framework streamlines the repetitive parts of grant consulting, giving me more time for what has never been mechanical: listening to executive directors, understanding what a program means to the people it serves, and finding language that honors their experience and makes its impact clear.

I do not need certainty about AI’s future to use it thoughtfully today; I need values strong enough to guide its use—so that technology creates more room for human connection, discernment, and better storytelling, not less. 

Citations/Additional Reading

1. Virtuous and Fundraising.AI. “The 2026 Nonprofit AI Adoption Report.” 2026.https://virtuous.org/blog/2026-nonprofit-ai-adoption-report/Use for: 92% of nonprofits using AI; 7% reporting major mission-related improvement; the “efficiency plateau”; 81% of organizations using AI individually and ad hoc; and 47% reporting no AI governance policy.

2. TechSoup and Tapp Network. “The State of AI in Nonprofits 2025: Benchmark Report on Adoption, Impact, and Trends.” 2025.https://page.techsoup.org/ai-benchmark-report-2025Use for: 85.6% of nonprofits exploring AI, 24% having a formal AI strategy, and 24.6% already using AI for grant writing.

3. Candid. “Where Do Foundations Stand on AI-Generated Grant Applications?” August 14, 2024.https://candid.org/blogs/funders-insights-on-ai-generated-grant-application-proposals/Use for: the 2024 foundation survey finding that 23% would not accept AI-generated proposals, 67% were undecided, and 10% would accept them. Note in the post that this remains the latest comparable Candid data, rather than presenting it as a 2026 survey.

4. NonProfit PRO. “Report: How Artificial Intelligence Is Changing the Nonprofit Sector.” February 3, 2025.https://www.nonprofitpro.com/article/2025-ai-benchmark-report-how-artificial-intelligence-is-changing-the-nonprofit-sector/Use for: the 60% interest in AI for grant writing and fundraising optimization, as well as the 24.6% grant-writing-use statistic.

5. Julep CRM. “From Curiosity to Commitment: How Nonprofits Are Using AI in 2026 and What Has Changed in the Last Year.” February 24, 2026.https://www.julepcrm.com/blog/22026-from-curiosity-to-commitment-how-nonprofits-are-using-ai-in-2026-and-what-has-changed-in-the-last-yearUse for: common chatbot uses: grammar, spelling, and punctuation checks; brainstorming headlines and subject lines; and first-draft content development. This is a secondary source, so retain it as context rather than the sole authority for your central argument.

6. Instrumentl. “A Nonprofit’s Guide to Grant Prospect Research.” February 22, 2025.https://www.instrumentl.com/blog/grant-prospect-researchUse for: your discussion of prospecting as a strategic alignment process, not merely a list-building exercise.

7. Center for Effective Philanthropy. “AI With Purpose: How Foundations and Nonprofits Are Thinking About and Using Artificial Intelligence.” September 2025.https://cep.org/report-backpacks/ai-with-purpose-how-foundations-and-nonprofits-are-thinking-about-and-using-artificial-intelligence/Use for: the broader context that AI use is growing across the sector, alongside continuing concerns about security, accuracy, staff readiness, bias, and equitable implementation.

8. International Energy Agency. “Energy and AI.” 2026.https://www.iea.org/reports/energy-and-ai/executive-summaryUse for: the environmental-concerns paragraph. The IEA reports that data centers used about 415 TWh of electricity in 2024, or 1.5% of global electricity consumption, and projects their electricity use could more than double to about 945 TWh by 2030.

9. MIT News “Explained: Generative AI’s Environmental Impact” (January 17, 2025). This supports the general statement that generative AI has environmental consequences, including increased electricity demand and water consumption. It is a university-based explainer, not a data-center operator’s marketing piece. 

10. H2O Cooling — “Closed-Loop vs. Open-Loop Cooling Tower: Which to Choose?” (August 20, 2025)  

11. Goldman Sachs Research. “How Will AI Impact the Labor Market?” Goldman Sachs Exchanges, July 2, 2026. https://www.goldmansachs.com/insights/goldman-sachs-exchanges/how-will-ai-impact-the-labor-market