Plenty has already been written by pastors, for pastors, about whether using AI in sermon prep is spiritually appropriate. That’s a valuable conversation, and not one I’m best positioned to add to. I’m not a pastor. I’m a practicing attorney who has led an AI governance function for years, and what I bring is a different lens: the same legal and practical questions I’d ask about any new tool at work, translated into what actually happens when a sermon prep tool touches real people’s information and your own credibility.
Where AI is a legitimate research partner
There’s a meaningful difference between AI helping you prepare and AI doing the preparing for you. Summarizing historical background, surfacing cross-references, generating a rough outline to react against, finding illustration ideas, checking your draft for clarity: none of this is fundamentally different from what pastors have always done with commentaries and concordances. It’s faster, but the posture is the same.
Who owns what you generate
Most pastors have never read the terms of service of the AI tool they’re using at 11pm on a Saturday. Many consumer AI tools’ terms grant the provider a broad license to use, store, or retain what you input, and free tiers often reserve more rights than paid ones. If a sermon outline or turn of phrase came substantially from an AI tool under those terms, the ownership picture is murkier than “I wrote this.” For anything you plan to publish widely (a book chapter, a podcast script), that’s worth knowing before it becomes a bigger question.
As I mentioned in the first installment, if you’re already paying for Netflix and Spotify without a second thought, a paid AI subscription is the same category of expense. It typically buys meaningfully better terms and a training-data toggle the free version doesn’t offer.
Confidential pastoral information can leak in without you noticing
This is the one that should actually worry you most. Sermon illustrations often draw on real counseling conversations, even anonymized ones. Typing that into a free AI tool may feed real, identifiable pastoral care information into a system whose data handling you haven’t reviewed. The same standard from piece one should apply to your own laptop, not just the church’s official tools.
A policy that only applies to staff and volunteers, but not to the pastor’s own habits, is difficult to defend if it’s ever actually tested.
AI can put words in a real person’s mouth
AI tools sometimes generate confident, fabricated quotes and attribute them to real theologians or historical figures, a different risk than a bad paraphrase. Repeating that from a pulpit is a real reputational and integrity exposure. The fix: never repeat an AI-generated quote publicly without verifying it against a primary source first.
BEFORE YOU USE ANYTHING AI-GENERATED IN A SERMON
Ask: “Could I stand behind this if someone asked me where it came from and what I believe about it?” If the honest answer is “I’m not sure, I just liked how it sounded,” slow down, verify it, and make it your own conviction, not just borrowed language.
The simple version
None of this requires avoiding AI in sermon prep. It requires treating it the way you’d treat any other tool that touches real people’s information and your own credibility: know what happens to what you type in, verify anything you didn’t originally source yourself, and hold your own habits to the same standard you’d ask your staff to meet.