Here is why, and how to write one
Your parish probably already has AI in its communications work. You may not know about all of it.
Someone has used ChatGPT to tidy up a notice for the bulletin. The priest has asked an AI tool to summarise a long diocesan document so he could read it before a meeting. A volunteer has used an AI image generator to produce a picture for a social media post. A youth minister has tried, quietly, having an AI draft a short reflection on the Sunday Gospel, then rewritten parts of it. A catechist has asked an AI for ideas for a First Communion activity sheet.
None of these are necessarily wrong. Some of them are fine; some may be useful; some are on harder ground than the people doing them have thought about yet. The difficulty is not any one of these examples. The difficulty is that the parish as a whole has not sat down and had the conversation. There is no shared understanding. There are no agreed boundaries. There is no way for the parish secretary to say to someone, “that is not what we do here,” because nobody has written down what we do here.
A policy fixes this. Not by banning AI. By giving the parish language for its own position.
The guiding principle
One sentence, which does most of the work:
The human is the oversight in sensitive, spiritually driven communication.
Not AI is bad. Not humans are better than machines. The human is the oversight. AI can accelerate the human’s work. AI cannot replace the human’s discernment, judgment, pastoral responsibility, or priestly accountability. That line, held clearly, resolves most of the practical questions a parish will meet in the next five years of these tools.
The principle is not new. The Church has thought for a long time about the difference between instruments and agents. A microphone is an instrument; the preacher is the agent. A printing press is an instrument; the editor is the agent. A translation service is an instrument; the translator who signs off on the final text is the agent. AI is an instrument. Someone is still the agent, and that someone is answerable, in the way human beings are answerable and tools are not.
When a parish loses hold of this distinction, the problems begin. When it holds the distinction firmly, AI becomes useful without becoming troubling.
Where AI belongs
Concrete examples of legitimate tool-use, where AI accelerates human work rather than replacing human discernment:
- Cleaning up grammar and typos in a newsletter draft the curator has written.
- Translating an operational notice into the three languages spoken in the parish, with a speaker of each language checking the result before publication.
- Summarising the bishop’s fifteen-page pastoral letter so the parish secretary can read the summary and decide which sections need highlighting in a two-paragraph version.
- Transcribing the notes from a PPC meeting, where everyone present has consented to the recording.
- Suggesting an alternative phrase when the curator is stuck on a sentence, which the curator may or may not use.
- Drafting the first version of a routine, non-pastoral announcement (a car park closure, a hall booking change) which the curator then reviews and shapes.
- Producing alt-text for images on the parish website, reviewed for accuracy before publication.
None of these threaten the guiding principle. In each, the human remains the author, the editor, and the one accountable for what is published. AI is doing what a very fast, rather literal assistant would do; the human is still signing the letter.
Where AI does not belong
Also concrete examples. These are not edge cases; they are the substance of pastoral communication, and the places where the principle matters most:
- A homily generated by AI on Saturday evening, with the priest having done no prior work on the readings. The homily belongs to the priest’s own prayer, study, and sacramental ministry. It is not a content problem to be solved by automation.
- A condolence message generated from a template and sent to a bereaved family. What the family needs is presence; what they receive, if the message is AI-written, is absence pretending to be presence.
- A weekly reflection posted under the priest’s name that he did not actually write, nor meaningfully revise. This is a problem of honesty, not of technology.
- A response to a pastoral enquiry from a parishioner in spiritual distress. No AI has the formation to discern such a situation. The parish secretary or priest does, however haltingly.
- Any communication where the reader would reasonably feel deceived to learn AI had produced it. The test is the reader’s reasonable expectation. If the reader would feel the message was something other than what it turned out to be, the use is wrong.
The line is not crossed at the technology. It is crossed at the point where discernment is absent and a reader has been led to think a human was present when a human was not.
The specific worry: sacred imagery
A particular case worth naming, because it is coming up quickly. AI-generated images of Christ, of Mary, of particular saints, of the Eucharist.
The Church has thought for centuries about sacred imagery. Through the theology of icons, through the traditions of religious art, through canon law on sacred objects, she has developed a careful sense that sacred images are not simply decorative but are a form of speech about the sacred, undertaken by people whose hands and prayer are part of what they produce.
An AI-generated Sacred Heart, produced in thirty seconds from a text prompt, with no human hand and no human prayer involved in its making, is not obviously the same kind of artefact as a commissioned icon or a photograph of a commissioned statue. The Church has not yet given final guidance on this, and practice will develop over the coming years.
The prudent default, while the Church is thinking, is traditional imagery, commissioned human work, or carefully chosen photographs of existing sacred art. This is not a rejection of AI. It is simply the recognition that sacred imagery is a more considered category than a social media graphic, and that the parish’s first experiments with image generation should not be with the things that matter most.
How to write the policy
Six steps. None of them are hard.
Step one. Download the True Light Digital AI Use Policy Template. It is the T8 resource in the Formation library. It is free, editable, and designed to be adapted rather than adopted verbatim.
Step two. Call a specific meeting. The priest, the parish secretary, and ideally one PPC member. An hour is enough. Put it in the diary as what it is: a conversation about how the parish handles a new technology.
Step three. Work through the template together. Adapt the pre-populated lists of permitted and non-permitted uses to your parish’s specific situation. Add cases the template did not anticipate. Remove anything that does not apply to you.
Step four. Sign it. Date it. Note the date of the next review (twelve months is right; the tools change that fast).
Step five. Tell the parish. A short paragraph in the newsletter, a quiet note to any volunteer who has been using AI tools, a mention at the next PPC meeting. The policy is not a secret document; it is a public statement of how the parish is approaching a new technology. Being public about it is part of its value.
Step six. Revisit annually. The tools will change. Your understanding will have grown. A policy that was right in 2026 may need adjustment in 2027. Build the review into the parish’s annual rhythm, perhaps at the start of a new Ordinary Time.
Closing
The parish that has an AI policy is not the parish that has solved AI. It is the parish that has thought about AI, named its position, and created a shared foundation for the further conversations that will come. That is enough. More than enough.
The absence of a policy is what creates most of the problems. The presence of a reasonable policy, even an imperfect first draft, makes almost every subsequent decision easier. The parish secretary has something to point to. The priest has something to fall back on. The volunteer has something to read before she starts experimenting. The conversations stop being “what do we think about this?” and start being “here is what we said we would do; does this fit, or do we need to talk?”
Your parish will have this conversation eventually. You can have it deliberately, now, with the template and an hour, when nothing is on fire. Or you can have it reactively, later, when someone has done something the parish is uncomfortable with and nobody has the language to explain why.
The deliberate conversation is cheaper. It is also, frankly, more formative. An hour spent writing the policy will teach the three people in the room more about the principle than a year of reading articles about AI.
The template is on the website. The hour is available this month. The conversation is yours to call.
This piece is part of the Pillar 4 series within the True Light Digital Formation framework. For the full cornerstone essay on which it is based, see truelight.digital/formation/guardrails-and-discernment/.