Quick answer: A church AI policy answers four questions in writing: what AI may draft, what it may never touch, who approves content before it publishes, and how you handle Scripture, sensitive topics, and disclosure. A useful policy fits on one page, is written by three people in one meeting, and is reviewed once a year. This guide gives you the nine sections and sample wording for each.
Why the policy has to come first
Most churches adopt AI backwards. They pick a tool, start using it, and write rules later — usually after something awkward went out.
Reversing that order changes everything. When the boundaries exist first, the tool conversation gets shorter, staff stop guessing, volunteers know exactly where the line is, and your leadership stops feeling like technology is happening to them.
It also protects the people doing the work. A communicator with a written policy can say "our policy says a pastor approves that" instead of making a judgment call alone at 9pm.
Who writes it
Three people: a lead or executive pastor, whoever leads communications, and one elder or board member. That is enough authority to make it stick and few enough people to finish in one sitting.
Block sixty minutes. Bring the nine sections below. You will likely finish in forty.
Section 1: Purpose
One paragraph explaining why the policy exists. This frames everything that follows as stewardship rather than restriction.
Sample: "We use AI tools to help our team communicate more clearly and spend less time on repetitive work. This policy defines how we use these tools in a way that protects our doctrine, our people, and the trust of our congregation."
Section 2: Scope
Say who and what the policy covers. Be broad here — staff, volunteers, interns, contractors, and every tool, not just the one you bought.
Sample: "This policy applies to all staff, volunteers, and contractors using any AI tool for work connected to our church, including general chatbots, design tools with AI features, and our communications workspace."
Section 3: Approved uses
List what AI is welcome to do. Being explicit here keeps the policy from reading as a ban.
- Drafting event announcements, social posts, texts, bulletin blurbs, and newsletters.
- Rewriting or shortening existing approved copy.
- Brainstorming series titles, campaign ideas, and volunteer recruitment angles.
- Summarizing a completed sermon transcript into recaps, prayer guides, and study guides.
- Drafting ad copy, FAQ answers, and press releases for review.
Section 4: Prohibited uses
This is the section people actually remember. Keep it short and absolute.
- Pastoral care. Grief, crisis, hospital, counseling, and condolence messages are written by people.
- Conflict and discipline. Any communication about a dispute or a membership matter.
- New doctrinal positions. AI reflects what we already teach; it never establishes what we believe.
- Political endorsements. Candidates, parties, or ballot positions.
- Confidential information. Never paste prayer requests, counseling notes, giving records, or member data into any AI tool.
- Fabricated specifics. No invented statistics, testimonies, quotes, or outcomes.
That last bullet prevents the most common real-world problem: a plausible-sounding detail that nobody verified.
Section 5: Doctrinal boundaries
Point to what your church already affirms and say plainly that AI content must stay inside it.
Sample: "All AI-assisted content must align with our statement of faith. When a topic falls outside our stated positions, the assistant should defer rather than take a position, and a pastor decides how we address it."
In ChurchCopy, this is not just a document. Your statement of faith, values, and prohibited topics live in your AI Policy and Guardrails settings, and every prompt reviews them before responding — whether someone is using the AI Chat Assistant or generating a full Copy Package.
Section 6: Scripture handling
Three decisions, and they take about four minutes:
- Which translation do we use by default?
- May Scripture be paraphrased, or quoted only?
- Who verifies references before publishing?
Sample: "Scripture is quoted from our primary translation. Paraphrase is allowed in social copy only when clearly marked as a summary. Every reference is verified against the text by the person reviewing the content."
Section 7: Privacy and naming people
This section prevents the most painful mistakes.
Sample: "We do not enter personal information, prayer requests, counseling details, or giving records into AI tools. We do not publish anyone's name, photo, or story without their direct permission, and we never describe a person's crisis in public communication."
Section 8: Human review and approval
The non-negotiable. Make it specific enough to follow.
Sample: "No AI-assisted content is published without a named person reviewing it. Routine content is approved by the communications lead. Content published under a pastor's name, giving appeals, and vision communication are approved by that pastor. Approvals are recorded in our workspace."
Recording matters. When something goes wrong, the first question is always who approved it, and approvals that live on the copy answer that instantly.
Section 9: Disclosure
Decide once, then be consistent. Both of these are defensible:
- Quiet: treat AI like spell check — a tool, not something you announce.
- Open: a short line on your website explaining that staff use AI to draft routine communications and that everything is reviewed by a person.
What you should avoid is leaving it undecided, because then the answer depends on who gets asked.
Three levels, if you want a starting point
Churches generally land in one of three postures. Pick the one that matches your leadership's comfort and adjust from there.
Cautious: AI drafts internal and logistical content only. A pastor approves anything public. No AI use by volunteers.
Balanced: AI drafts most communications. Staff approve routine content, pastors approve anything doctrinal or under their name. Volunteers may draft inside the workspace but never publish.
Forward: AI drafts broadly including ads, blogs, and recaps. Approval thresholds by content type. Volunteers trained and given drafting access with staff approval.
Getting it approved
Bring three things to your board or elders and the meeting usually takes twenty minutes:
- The prohibited uses list. Lead with the boundaries, not the benefits.
- The review requirement, stated plainly.
- The reason: hours recovered, and what those hours will go toward.
Leading with efficiency makes leaders nervous. Leading with boundaries makes efficiency sound like stewardship.
Keep it alive
Put an annual review on the calendar the day you approve it. Ask three questions: has anything published that should not have, are the prohibited topics still the right ones, and is the human review step actually happening or has it become a formality?
A policy nobody checks becomes decoration. A policy reviewed once a year becomes the thing that lets your church use powerful tools without losing its voice.
A one-page policy you can adapt today
If you want a starting draft rather than a framework, copy this and change the bracketed parts. Most churches need fewer edits than they expect.
Purpose. We use AI tools to help our team communicate clearly and spend less time on repetitive work, in a way that protects our doctrine, our people, and the trust of our congregation.
Scope. This applies to all staff, volunteers, interns, and contractors using any AI tool for church work.
Approved uses. Drafting announcements, social posts, texts, bulletin copy, newsletters, email campaigns, ads, FAQ answers, volunteer recruitment copy, and summaries of completed sermons. Editing and shortening approved content. Brainstorming ideas and titles.
Prohibited uses. Pastoral care, grief, and crisis communication. Conflict and church discipline. New doctrinal positions. Political endorsements. Any confidential information including prayer requests, counseling notes, member data, and giving records. Invented statistics, testimonies, or quotes.
Doctrine. All AI-assisted content must align with our statement of faith. Where we have not stated a position, the assistant defers and a pastor decides.
Scripture. Quoted from [translation]. Paraphrase permitted in short-form copy only when clearly a summary. Every reference verified by the reviewer.
Privacy. No personal information is entered into any AI tool. No one is named, pictured, or described without direct permission.
Review. Nothing publishes without a named person approving it. Routine content is approved by [role]. Content under a pastor's name, giving appeals, and vision communication are approved by [role]. Approvals are recorded in our workspace.
Disclosure. [We treat AI as a drafting tool and do not disclose routine use / We publish a statement that staff use AI for routine communications with human review.] We never use AI for personal or pastoral content.
Review date. This policy is reviewed annually in [month].
Ten paragraphs. Bring it to a sixty-minute meeting with three people and you will walk out with something your board can approve.
Try for free — load your policy into your workspace and see it applied to every draft.