Quick answer: Set your primary Bible translation in your AI settings, decide whether paraphrase is allowed, and require a person to verify every reference against the text. On disclosure, pick a posture and be consistent — either treat AI as a drafting tool like spell check, or publish a short statement that staff use AI for routine communications with human review. The failure is leaving it undecided so the answer changes depending on who gets asked.
Two questions that keep coming up
Once a church starts using AI seriously, two questions surface within the first month. Both are about honesty, which is why they land harder than the technical questions.
How do we make sure Scripture is handled correctly? And do we have to tell people?
Neither is complicated. Both need an actual decision rather than a shrug.
Part one: Scripture
The three failure modes
Wrong reference. The quoted text is accurate but the citation points elsewhere. This is the most common error and the easiest to miss, because the verse itself reads fine.
Translation drift. Your church teaches from one translation and your social posts quietly use three others. No single post is wrong; the cumulative effect is inconsistency your regular attenders will notice.
Paraphrase that shifts meaning. A tightened, more quotable version of a verse that has moved a step away from what the text actually says. This is the one with real theological consequence.
The three decisions
- Primary translation. Set it once in your settings so every draft defaults to it. This eliminates drift entirely.
- Paraphrase policy. Many churches allow paraphrase in social copy when it is clearly a summary, and require direct quotation anywhere Scripture is being taught. Pick a line and write it down.
- Verification owner. Whoever reviews the piece checks the references. Not "someone." The reviewer.
Those three decisions take about five minutes and remove nearly all Scripture risk. Put them in your AI policy settings so they apply automatically instead of depending on memory.
A note on Scripture in ad copy
Short-form ad formats compress everything, and Scripture compresses badly. A verse cut to fit a headline can end up meaning something the passage does not say. The safest rule is to paraphrase the idea in your own words rather than trimming the text itself.
Part two: disclosure
Why the question feels heavier at church
Nobody expects a company to announce that AI drafted its product email. Churches are different, because churches ask people to trust them about things that matter far more than a product.
So the instinct to ask "are we being honest?" is a good instinct. It just needs to be pointed at the right question.
The two defensible postures
The tool posture. AI is a drafting aid like spell check, templates, or stock photography. Nobody discloses those. A person wrote the brief, a person reviewed the output, a person approved it, and that person stands behind it. No announcement needed.
The open posture. A short line on your website or in your handbook: "Our staff use AI tools to help draft routine communications. Every message is reviewed and approved by a member of our team before it is published." Clear, brief, and it removes any possibility of a gotcha later.
Both are honest. Choose based on your congregation's culture, not on what other churches do.
The line that is not optional
Where disclosure stops being a preference and becomes an integrity issue: anything presented as one person's individual words to another person.
- A pastor's personal note to a family.
- A condolence or encouragement message to an individual.
- A personal testimony or first-person story.
- A leader's statement on a hard subject.
If the recipient would feel misled knowing AI drafted it, AI should not have drafted it. That is not a disclosure question — it is a boundary question, and it belongs in your prohibited list alongside pastoral care.
What about sermons?
Churches are still working this out, and there is genuine disagreement. The common landing spot: research and outline assistance is treated like commentary use, while the sermon itself is written and owned by the preacher. Whatever your church decides, decide it explicitly rather than letting it be assumed.
Answering the question when someone asks
Eventually a member will ask directly. Usually casually, in the lobby.
The answer that works is short and unembarrassed: "Yes, our team uses AI to help draft routine announcements and newsletters, which frees up time for the work that needs a person. Everything gets reviewed and approved by staff before it goes out, and we never use it for anything pastoral."
Three sentences. Confident, not defensive. Most people nod and move on, because the concern was never the technology — it was whether anyone was paying attention.
Write both answers down
Add a short section to your policy covering both halves:
"Scripture is quoted from [your translation]. Paraphrase is permitted in short-form copy only when clearly presented as a summary. Every reference is verified by the reviewer before publishing. Our church [does / does not] publicly disclose AI assistance in routine communications. We never use AI for personal, pastoral, or first-person content."
Four sentences, and both questions are permanently settled for your whole team.
Why this matters more than it seems
Scripture handling and disclosure are the two places where AI use at church touches integrity rather than efficiency. They are also the two places leaders are most likely to postpone deciding, because neither feels urgent until it does.
Deciding both now costs ten minutes. Deciding them after a member raises a concern costs considerably more, and the conversation starts from a much worse position.
Churches that handle these two well end up with something valuable: the ability to use powerful tools openly, without anyone wondering what is being hidden.
A five-minute Scripture audit
Before you settle your policy, it is worth finding out how your church currently handles Scripture in practice. The gap between what you assume and what is actually published tends to surprise people.
Pull the last twenty pieces of communication that quoted Scripture — social posts, newsletters, bulletins, graphics. Then check three columns.
Translation. How many different translations appear? Most churches find at least three, and nobody chose them. They came from whatever source was closest at the moment.
Reference accuracy. Verify each citation against the text. Errors here are more common than teams expect, and they long predate AI — the difference is that AI produces more content, so a low error rate spread over more pieces means more total errors unless verification is deliberate.
Paraphrase. Mark anything that is a summary rather than a quotation. Then ask whether a reader would know the difference. If a paraphrase is formatted in quotation marks with a reference, it is being presented as text it is not.
That audit takes about five minutes and it usually settles the policy conversation on its own. Teams that were inclined to treat Scripture handling as an unnecessary formality tend to change their minds once they see three translations and two misfiled references in their own recent work.
Then set the translation in your settings, write the paraphrase rule, assign verification to the reviewer, and the problem is permanently solved for everything you publish going forward.
Try for free — set your translation and boundaries once, applied to every draft.
