Signal Check: AI in the NFP Sector

The goal of this newsletter has been to help translate what’s happening in AI into something useful for those of us working in NFPs.

Some weeks that means going deep on one idea or topic. This week, I saw enough movement in the news that spoke directly to what I am working on leading AI adoption at my own organization that I wanted to share.

Issue 3 is a roundup: five things worth knowing, what each one actually means for NFPs, and where I'd point your attention.

That's not a typo. The 2026 Nonprofit AI Adoption Report from Virtuous surveyed 346 organizations and found near-universal introduction of AI tools paired with near-invisible fundraising results. Nearly half of respondents said they have no formal AI strategy or repeatable workflows.

The reflex read is "everyone needs a strategy”, but as you can imagine, I'd push back. Near-universal adoption means the sector is, by and large, getting reps in; drafting emails, asking LLMs questions, experimenting with content creation. The 7% impact number isn't a failure of planning; it’s proof that off-the-shelf introduction of AI tools, and battle-hardened adoption frameworks for AI fundraising, do not yet exist.

The 7% are ready and able to move forward intentionally with confidence drawn from foundational experience.

A recent WordPress VIP survey dropped this stat and it should give every fundraiser and marcomms team pause. If you're experimenting with with "AI-powered" donor communications, its important to be thinking about where in the process is AI actually showing up.

This reinforces the position I generally subscribe to for most applications: AI helps expertise do more, and compliments your team’s work rather than replaces it end-to-end. A communications team can use AI to help build out more variations of output, design more A/B tests, or help research more hyper-personalized messaging. None of that needs to mean the final drafts aren't written, or at least edited, by the experts you've trusted your organization's voice to.

The CEP report fills in the picture the adoption numbers don't capture. Yes, 92% are reporting use of AI. But how many are thinking about who gets the benefits? The survey found that equity considerations are lagging behind adoption — meaning the organizations deploying AI the fastest aren't necessarily the ones thinking hardest about fairness, bias, or access.

Worth sitting with: equity instincts in AI don't develop in the abstract. They develop when people in your org are close enough to the tools to see where bias shows up, where access breaks down, where outputs go sideways. Foundations and NFPs waiting for a perfect equity framework before engaging will be likely be the last to develop real judgment.

Gates Grand Challenges is funding projects that use AI to help donors give more, and give sooner. The grants are modest by foundation standards, but the signal is what interested me the most rather than the dollar amount.

One of the largest funders in the world is putting money behind AI for good in our specific corner of the sector. Many of the AI features bolted onto the fundraising and CRM tools you already use can feel underwhelming; like a checkbox rather than a step-change. Targeted funding for purpose-built tools is exactly what this space needs.

Leaked financials put OpenAI's loss at $38.5 billion. When people talk “AI bubble” it’s not because there’s no value in these tools today or where they are projected to head, it’s that the costs are outstripping any semblance of economic viability unless further innovation delivers and does so quickly.

For me, this is yet another signal that supports the at-bats thesis. Getting exposure to the technology, how it works, where it is strong and weak, is the most valuable approach while the tooling landscape is still shifting and unsettled.

It also speaks to why it’s important to be intentional about what is being shared with AI tools. We live in a digital age where many products are free or subsidized to use, because the economic model focuses on what it can extract from the user and what is provided to it to harvest.

For NFPs, that means applying the same judgment you would to any external system: don’t share what you wouldn’t be comfortable storing, retaining, or seeing outside your control.

That's it for this week. See you next Tuesday. If you enjoyed this type of content or preferred the longer form strategy pieces, I’d love to hear either way.