The Ground Is Shifting. Do More Than Take Notes
By Brett Horvath and Daphne Davis Moore
We’re in a conference room at ComNet 2025 with senior leaders, discussing how to integrate AI, specifically AI agents, not only for creating content but also to build strategy, find alignment, and surface bias.
At first glance, we’re an odd pair. I’m sporting a corporate bob with Oprah’s favorite readers perched on top. Brett rolls in wearing a funky suit, an embroidered hat, and sparkly nail polish that catches the light. We don’t match on the surface, but beneath the readers and the hat, you’ll find brains and hearts aligned.
Brains that think differently, but in the same direction. Hearts that beat differently. We both have deep roots in rural America. A shared belief that technology used wisely helps nonprofit organizations hit their goals more effectively today and transform for the future with efficiency, effectiveness and excellence. But, and most importantly, a growing conviction that the sector is having the wrong conversation about AI.
Or at least an incomplete one.
That was a more than six months ago, but also light-years considering how AI has compounded its capabilities since then, making what was an important conversation then an urgent one today.
A Sector Under Pressure
According to research from the Center for Effective Philanthropy, many nonprofits are facing what leaders describe as existential threats. The converging headwinds are not abstract: tightening budgets, shifting political dynamics, evolving rules and norms, and teams stretched past their breaking point. These are problems that demand new forms of judgment and organizational infrastructure, not better software. And they are arriving at a pace that outstrips most strategic plans.
AI doesn’t address those problems by making a newsletter more efficient. It addresses them by transforming how organizations make decisions, understand their stakeholders, and respond to a world that is changing faster than any three-year plan can track.
In the face of existential threats, nonprofits need more than band-aids; they need a quantum leap forward, and we have the tools today to take that leap.
That is the conversation the social sector needs to have — and largely isn’t.
Dabbling Is Not Transforming
The data tells a clear story about where the sector stands.
According to a 2026 nonprofit AI adoption study, 92% of nonprofits now use AI in some capacity, but only 7% report major strategic impact, and just 4% have documented, repeatable workflows. Three-quarters of nonprofits lack an AI strategy, according to Tech Soup and the TAPP network. And the gap between funders and the organizations they support may be the most revealing indicator: the Center for Effective Philanthropy reports that 87% of nonprofits say foundation staff doesn’t understand their AI needs, while 93% of foundation leaders believe they do.
The pattern is consistent: adoption is nearly universal, but capacity is almost nowhere. Organizations are using AI widely and building with it almost not at all.
The pace problem is structural, not just cultural. Among the minority of organizations that have adopted formal AI policies, the recommended review cadence is every six months. The technology is fundamentally shifting every two to three months. In 2025 alone, major AI labs released new foundation models roughly every eight to twelve weeks, each generation introducing capabilities that rendered previous governance assumptions incomplete. By the time a policy is reviewed, two or three generations of new models have already shipped. Organizations are governing yesterday’s technology with last year’s framework. And the 70 to 90 percent of organizations that have no policy at all are not yet in the conversation, not because they don’t care, but because the infrastructure to help them get there doesn’t exist.
The most common uses of AI across the sector? Meeting transcription and drafting emails.
We have access to transformative technology, and we are using it to take notes.
This is dabbling, not transforming. And the gap between the two has consequences.
Some will argue, and not without reason, that AI is itself a distraction from the real work. That the social sector’s problem isn’t insufficient technology but insufficient power: underfunded organizations, eroded civic infrastructure, and a political environment hostile to the communities nonprofits serve. That every dollar and hour spent on AI adoption is a dollar and hour not spent on organizing, advocacy, or direct service. That they’ve been sold transformative tools before, and the transformation never arrived.
That argument deserves to be taken seriously, not dismissed. And it would be right if AI were simply another tool to layer on top of existing operations. But the shift we’re describing is not about another grants management platform. It’s about whether organizations can keep pace with the environment they’re already operating in — an environment where the threats to democratic participation are already AI-accelerated, whether the sector engages or not.
The actors who are not waiting are specific and identifiable. Criminal networks are using AI to target financial information, forge identities, and run personalized scams at a scale no human team could match. Deepfake-driven fraud alone caused over $200 million in losses in the first quarter of 2025. Influence operations are producing tailored disinformation designed to erode voter confidence, public trust in public health, and institutional legitimacy. Predatory businesses are deploying AI to exploit consumers, particularly seniors and economically vulnerable populations, with practices that are harder to detect and faster to scale. And the collapse of local news has left more than 1,700 U.S. counties with limited or no local coverage, affecting some 50 million Americans and creating information vacuums that AI-generated content is already filling.
These risks compound each other. And the emergent safety challenges from AI itself, for which there is currently no meaningful regulation or institutional infrastructure to respond, are accelerating all of them.
We raise these points not to alarm but to clarify. In previous technology cycles, organizations could afford to wait: commission research, benchmark, run pilots, let the landscape settle. AI is not settling. And the organizations and communities the social sector serves are already living inside these dynamics, whether the sector has caught up or not.
The compounding consequences of slow adoption extend beyond inefficiency. They are creating a widening capability gap at the precise moment when mission-driven organizations can least afford one.
Recently, Mike Kubzansky, departing after eight years as CEO of Omidyar Network, argued that philanthropy must “move faster, go deeper, and engage more directly with AI and its ramifications.” Allison Scott, CEO of the Kapor Foundation, has written that the “collective influence of philanthropy can begin to shift capital, knowledge, and power toward AI innovation and usage that works for the common good.” These are signals that the sector’s leadership recognizes what the data confirms: the next move could be collective.
The Partnership This Moment Requires
The 87% statistic from the Center for Effective Philanthropy, the share of nonprofits reporting that funders don’t fully understand their AI needs, is worth sitting with. Not as an indictment, but as a signal about what this moment asks of the funder-grantee relationship.
Funders and grantees are approaching AI from different positions, and those positions are genuinely complementary. Foundations bring institutional resources, longer planning horizons, and the capacity to absorb risk. They have rightly oriented attention toward consequential questions: bias, energy use, community impact. Nonprofits bring proximity to communities, operational urgency, and firsthand knowledge of what capacity gaps actually look like on the ground.
Right now, these strengths largely operate in parallel. Nearly 90% of foundations provide no AI implementation support to grantees. That means funders are missing the ground-level intelligence that would sharpen their AI strategy, and grantees are missing the institutional backing that would accelerate their capacity.
Closing that gap would look like what the best funder-grantee relationships already aspire to: honest exchange, mutual investment, and a shared stake in outcomes. The difference is that AI’s pace makes the usual timeline for finding alignment untenable. This partnership has to move at the speed of the challenge.
What Higher-Level Use Actually Requires
Moving from experimentation to transformation requires being precise about what transformation means.
Transformation means structuring the environment so that human judgment operates faster, with fewer blind spots, and with greater alignment across teams and stakeholders. Not replacing human judgment. Strengthening and extending it.
We call this principles-based computing: an approach in which organizations define their core operating principles and then structure AI to reason from those principles, not just execute tasks. Rather than asking “what can this tool do?”, it asks “what does our organization need to decide, and how can AI extend our judgment while keeping our values intact?” It is the difference between automating what you already do and transforming how you think.
None of this requires large budgets or technical staff. It requires a different set of questions.
A Moonshot Worth Naming
The goal is ambitious but achievable: to move the sector from patchy experimentation to shared capacity at scale. Not AI as a program area. Not AI as an add-on. AI as upgraded infrastructure for judgment, decision-making, and collective action under pressure.
This is not without precedent. The interstate highway system began as a military logistics project and transformed civilian commerce. The World Wide Web was built on a simple, principles-based agreement about format, and that agreement unlocked a transformation in how the world shares information. Each of these began with the same insight: when powerful resources are locked up in a few institutions, the solution is to build the infrastructure that lets them share.
Today, the AI equivalent of those supercomputers is concentrated in a handful of companies. The social sector doesn’t need to build its own large language models. What it needs is its own internet moment: shared infrastructure that lets mission-driven organizations access, learn from, and build on each other’s judgment at a speed that matches the environment they’re operating in.
The social sector has, at this moment, both the need and the opportunity to move beyond the familiar conversation. The ground is shifting. It’s time to do more than take notes.
About This Series
This is the first in a series examining what becomes possible when the social sector treats AI not as a tool to adopt, but as a force already reshaping the landscape it operates in.
Subsequent pieces will take up:
How AI is shifting your audiences;
The risk that time itself poses to the sector;
Why AI is not a program area but is already transforming every issue in the portfolio from climate to democracy to health;
What higher-level use looks like when put to work with real organizations facing real decisions.