When Nonprofits and Churches Should and Shouldn't Use AI

We hear a lot today about artificial intelligence and how it is a game changer, but rarely do we hear people say when you should not be using it.

AI can help your organization do more things in less time. It can automate workflows and make parts of your organization more efficient. But that does not mean you should automate everything.

The question should not only be, "How can AI help with this?"

The better question is, "Should AI be helping with this?"

There is a lot of research saying that now is a good time for nonprofits to start paying attention to AI, and I agree with that. A 2025 TechSoup research report found that 85% of surveyed nonprofit organizations have high interest in generative AI and predictive analytics for their organizations.

That makes sense. Why should only massive technology companies benefit from the investments being made in artificial intelligence? If AI can help organizations care for their communities, strengthen communication, support donors, improve workflows, and make better decisions, then churches and nonprofits should be able to use those tools too.

But there are also real words of caution circulating through the nonprofit sector.

AI is developing faster than ever before, but what seems less clear is whether most organizations have a plan for how to use it well. TechSoup also reported that 76% of nonprofits say they do not have an AI strategy. Many organizations have one or two people looking into AI capabilities, while others are rightly worried about the social effects of these tools.

Those concerns are valid. There is no point in adopting powerful new technology if there is no clear plan for what it should do, what it should not do, and who is responsible for the outcomes.

TechSoup also reported that 47% of nonprofits only have one staff member making all IT and AI decisions. With limited staff, there is only so much one person can do to map out everything AI is capable of doing while also protecting the organization from mistakes. Only a small percentage of nonprofits outsource help with AI, but I think that is an area where more organizations should be open to support.

Unless you have people on staff who understand software, operations, data, workflow design, and implementation, it is risky to internalize every decision about AI without outside guidance.

I understand that the financial strain is real for nonprofits. Budget constraints are not just theoretical. But that is exactly why AI decisions need to be made carefully. Once AI is connected to real workflows, there can be costs tied to automation tools, model usage, subscriptions, storage, implementation, maintenance, and monitoring.

Poorly built workflows can waste money by sending unnecessary information to AI models or using AI where a simple automation would have worked just as well.

That is why the real question is not just, "Can we use AI?"

The bigger question is, "Where should we use it, where should we avoid it, and what system needs to be in place first?"

When You Should Be Using AI

Grant writing is one of the clearest areas where AI can make a real impact for nonprofits. AI can help with prospect research, project outlines, first drafts, statements of need, tone adjustments, proofreading, and compliance checks.

That does not mean AI should write everything without review. Sensitive data needs to be protected, statistics need to be fact-checked, and the final writing still needs a human voice. But when used carefully, AI can save real time and help teams focus more energy on the mission.

AI can also support content marketing. That includes communication for potential donors and volunteers, but also for current supporters. With human review, AI can help draft social posts, email campaigns, donor updates, event reminders, and other recurring communication. It can help connect messaging across channels without requiring staff to start from scratch every time.

Another area where AI can make an impact is strategy. Smaller nonprofits and churches usually do not have entire teams analyzing trends, donor behavior, program performance, staffing needs, and future opportunities.

AI can help leadership compare information, summarize reports, identify patterns, and think through different scenarios more quickly. It can help answer questions like which programs are growing, where staff time is being wasted, what areas are underperforming, and what leadership should be watching next.

The key word is help.

AI should support leadership decisions, not replace them.

The same principle applies to finance. AI can help categorize transactions, identify unusual spending, compare budgets to actual results, summarize financial reports, and assist with forecasting. Automated workflows can also route invoices, send approval reminders, and generate recurring reports.

But AI should not independently decide where money goes, whether an expense gets approved, or whether a major financial decision gets made.

Operations may be one of the best areas for churches and nonprofits to use AI and automation. A volunteer fills out an application, someone gets notified, a task is created, onboarding information is sent, and a reminder goes out if nothing happens. A donor inquiry comes in, gets categorized, is assigned to the right person, and is tracked. An event registration triggers a confirmation and reminder.

A lot of these tasks do not even need AI. A simple automated workflow may be better because it follows the same rules every time.

The goal is not to add AI everywhere. The goal is to remove repetitive work that does not need someone manually handling every step.

When You Shouldn't Use AI

One of the biggest mistakes organizations can make is confusing AI's ability to recommend something with its ability to make a final decision.

AI can analyze information, identify patterns, summarize data, and suggest options. That does not mean it should have the final say on decisions that affect real people.

I would be very cautious using AI to independently decide whether someone receives assistance, whether a volunteer is accepted, whether an employee should be terminated, how someone in crisis should be counseled, or how a major financial decision should be handled.

Churches and nonprofits deal with people who may be sharing financial problems, personal situations, health information, prayer requests, donor information, and other sensitive details.

Efficiency is not worth damaging trust.

Organizations also need to be careful with open-ended AI systems. There is a major difference between giving AI one clearly defined job and giving it broad access to everything.

For most organizational workflows, I strongly prefer closed-loop AI. AI should have a defined task, limited access to information, clear rules about what it can do, and human review before anything important happens.

For example, AI can summarize a donor inquiry, categorize it, and draft a response. A staff member can then review and send it. That is much safer than giving AI access to email, financial records, donor information, internal documents, and other systems and simply telling it to handle whatever it thinks needs attention.

The more open-ended the system becomes, the more chances there are for mistakes, bad assumptions, privacy problems, or actions that should never have happened.

AI is also almost useless if the systems underneath it are bad. If information is spread across spreadsheets, personal drives, outdated databases, email inboxes, and software nobody updates, AI is going to struggle to produce reliable answers.

Before adding AI, an organization should know where its information lives, what its source of truth is, who owns each process, what AI is allowed to access, and what still requires human approval.

This also matters financially. Once AI is integrated into real workflows, costs can increase quickly if the system is poorly designed. Running AI when a basic automation would work is wasteful. Feeding messy or unnecessary data into a model is wasteful. Automating a broken process only makes the broken process move faster.

The goal should never be to use as much AI as possible.

Sometimes the right answer is AI. Sometimes it is basic automation. Sometimes it is fixing the process. Sometimes it is simply having a person do the work.

The best question to ask is what happens with the time AI saves. If AI saves three hours of repetitive work, can those three hours go toward serving people, building donor relationships, helping volunteers, improving programs, or moving the mission forward?

That is where AI becomes valuable.

The churches and nonprofits that use AI best probably will not be the ones that automate the most. They will be the ones that understand exactly what should be automated, what should be assisted, and what should always remain human.