Does Your Church Need an AI Policy?

Short answer: Yes, and it should fit on one page. Your staff are already using AI whether you have a policy or not. Saddleback's policy comes down to five guidelines: check the theology, keep a human in the loop, be transparent when it matters, never paste in member data, and remember technology never replaces human interaction.

I am the Innovative Tech Pastor at Saddleback Church. When we rolled Google Gemini out to our staff, we did not start with a rulebook. We started with a document that explained how to think about AI in our context, gave the five guardrails below, and showed practical prompts. Then we trained on it. Everything on this page is what we actually use, and the documents are free to download at the bottom.

What should a church AI policy include?

Ours has five points. I will give you each one the way we give it to staff.

1. Understand AI bias, especially around theology and cultural topics. Every language model is trained on data, and that data shapes the results. It is like a teenager who has only read ten books and thinks whatever those ten books say is true. AI models will not align with our theological convictions, so nobody copies and pastes a statement without checking it against what our church believes.

2. Human oversight is essential. AI can make up facts and data points. They call it hallucination. This is not so different from humans, who sometimes say something is true because it feels true. AI is an excellent resource for brainstorming, but you double and triple-check everything before it goes in a post, an email, or a public forum.

3. Be transparent when appropriate. AI content can carry an uncanniness; part of your brain recognizes something is off. It tends to be too long and fluffy, like a college student padding a paper. Shorten it and read it out loud. If you used AI to brainstorm, you do not need to disclose that. If you copy and paste and send it, you might. Better yet, alter the content enough that it is genuinely your own.

4. Protect member data. What you type into an AI tool can be used to train it, so your data is generally not private. Never drop birthdates, addresses, or full names of attendees and members into an AI model. This is also why our staff log in with their staff account for all church work rather than a personal one.

5. Technology never replaces human interaction. We leverage the tools at our disposal in the time we exist. Just like a fire can burn your house down or warm a good cup of coffee, we discern how we use the tools of our day. It is never about being relevant. It is about being effective. We will always physically gather and talk to people face to face while using technology like AI in the process, if it helps our ministry objectives.

Can pastors use ChatGPT or AI for sermons?

For brainstorming, research, finding the right verse, and engaging a tough topic from several angles, yes. For writing the sermon, no. The theological bias problem in guideline one is most dangerous exactly here, because AI output sounds authoritative and gets doctrine wrong with total confidence. Use it the way you would use a well-read friend who is not a member of your church: helpful for ideas, never the final word, and never copied into the pulpit.

On disclosure, the same rule applies as everywhere else. If AI helped you think, you do not owe anyone an announcement. If you are reading its words, your congregation deserves to know, and honestly you should not be reading its words.

How do you get a church staff to actually use AI?

You do not hand them a perfect, finished solution. You talk about it out loud. Here is the shift that changed things for us: I stopped trying to be the person with all the answers on AI and started being the person who models curiosity about it. I use it imperfectly and publicly, and the prayer is that it unlocks something for the people watching.

The goal is not to make every staff member an AI expert. The goal is to get them problem-solving. The moment a children's ministry director or a facilities manager asks, "Wait, could this help me with that thing I do every week?" you have won.

Our staff training ran in five parts:

1. How to think about AI as a staff member. Mindset before mechanics. What it is good at, what it is not, and how to treat it as a helper rather than a threat. Focus on your job, not the task.
2. Beginner ways to use it. Rewrite an email in a softer tone. Summarize a long thread and draft a reply, then review before sending. Brainstorm ideas for a project with the audience and outcome specified.
3. Immediate ways to use it. Run a premortem on an event. Prep for a meeting. Summarize meeting notes in a short, scannable format.
4. Advanced ways to use it. Where it goes once the fear is gone. Persona reviews, custom assistants for repeated tasks, and eventually agents working inside your systems.
5. Staff share how AI makes them more productive. Real examples, in their own words. This did more to inspire the room than anything I said from the front.

That last part is the secret. Experts on a stage inspire admiration. Coworkers inspire action.

We are a Google shop, so our training leans on the tools our staff already open every day. If your church runs on Microsoft, swap the names and the principles carry over cleanly.

What can AI actually do inside a church's systems?

More than most pastors expect, with one caveat I will get to. Since we moved to Rock RMS, I have used Claude to tweak our homepage, build an engagement timeline on people's profiles so staff can see how connected someone is before following up, create reports for forms, events, deletions, and retention, and diagnose bugs. One example: baptism data was not populating, and Claude flagged that a scheduled job had not run in a week. Another time it was wrong on the first few passes and then surfaced a rule in our jobs that was causing a tracking gap.

The caveat: this works because I deeply understand how our data is structured. Real developers were still involved and vetted my changes. As Andrej Karpathy put it, you can outsource your thinking, but you cannot outsource your understanding. Claude is the leverage. The understanding is the moat.

Where do I get the documents?

• Sample Church AI Policy and Training You Can Copy (the Saddleback AI overview as a Doc and a PDF)
• Give Your Team Permission to Explore AI (Here's Our Actual Training) (raw notes, keynote PDF, and the slide file)
• What I've Built in Rock RMS Using Claude
• How to Upgrade Your Ministry Team with AI
• 40 Bible Passages About AI