What Happened
A California pastor named Justin Lester has built a digital twin of himself, trained on nearly 2 million words taken from his sermons, books, and other messages. The artificial intelligence–generated clone offers counsel around the clock to hundreds of people in Vallejo, a city in the Bay Area. The New York Times, which reported on Lester, said the AI twin talks in the pastor’s cadence.
The clone’s record is concrete. It comforted a woman the night before heart surgery, advised a person who felt despair over a lost job, and generated a letter for a homesick child at camp. Lester is not alone in looking for ways to fold generative AI into Christian life.
One AI-infused app, Bible Chat, has built another form of digital chaplaincy. Users can confess their secrets and sins to a chatbot and ask it for insight about how to interpret the Bible. Bible Chat has been downloaded more than 30 million times, and other “faith tech” tools are also on the rise.
Christianity Today pushed back hard on the trend. Its argument: Christians should not be receiving pastoral guidance from technology that won’t give an account for their souls. The outlet then set aside the obvious objections with three questions: “Can a chatbot pray for you?” (No.) “Can it know you?” (No.) “Can it be trusted to tell you the truth when the truth costs something?” (No.)
The article’s sharper claim is about authority, not capability. When Lester created his digital twin, it wanted to extend his care to more people than one man can reach. According to the piece, what he did was curtail the one thing he was set apart to carry. Scripture describes the pastor’s task as keeping watch over souls as “those who will have to give an account” (Heb. 13:17).
What This Means For You
If you lead a congregation, the question is no longer whether AI touches your ministry. It is who signs off on it. Insist on a named human owner for every tool, from sermon prep to prayer-request triage to any chatbot wearing your church’s name.
Confession is the sharpest edge of this shift. If a member tells a chatbot about abuse, addiction, or self-harm, someone needs a written plan for what happens next. Chatbots do not file mandatory reports, do not sit with anyone in a hospital room, and generally do not carry the legal confidentiality that clergy do.
Think about where that disclosed data lives. A confession typed into an app now sits on a vendor’s servers, governed by that vendor’s retention policy and potentially reachable by subpoena. Before you recommend any tool, read the privacy terms yourself — not the marketing page.
If you are a layperson, watch what you are outsourcing. Asking a model to explain a difficult passage at 2 a.m. is convenient. It also trains you to expect instant answers without relationship, which is the opposite of discipleship.
If you are considering building a digital twin, resolve the consent questions before launch, not after. Whose words trained the model? Can a congregant opt out of being modeled or counseled this way? What happens when the pastor retires, dies, or changes theological positions and the clone does not?
Prepare for the operational side too. Ask your insurer whether AI-delivered counseling falls under existing liability coverage. Draft a one-page policy that states plainly what AI may do, what it may never do, and who a distressed person should contact instead.
Finally, watch for the next milestones. Denominational guidance, disclosure norms for AI-generated spiritual content, and vendor certifications for crisis handling are the three signals that will tell you whether this stays a novelty or becomes infrastructure.
Why It Matters
Protestantism has always carried tension around spiritual authority, and this suggests AI pushes that tension further. A pastor’s authority rested partly on presence, accountability, and the willingness to be wrong in front of people who know them personally. A clone can be endlessly available and never accountable.
The pattern is not unique to religion, though. The same accountability gap — automation operating without a responsible human in the loop — shows up in enterprise business safety and governance, where companies are still learning that “the model decided” is not a defense.
This could mean that AI-assisted counseling eventually draws scrutiny similar to other high-stakes deployments, as the broader debate over AI regulation has already shown. When software touches vulnerable people at vulnerable moments, disclosure and liability rules tend to follow.
Key Takeaway
- Scale is the selling point and the risk: a twin built from nearly 2 million words reached hundreds of people, but none of them got a person who answers for their soul.
- The governance question is practical, not philosophical. Name a human owner, review data retention, and write an escalation path for crisis disclosures.
- Consent is the unaddressed gap in faith-tech buildouts — congregants rarely know their words or confessions may train future models.
- Expect denominational and insurance guidance to arrive before regulators do, and treat vendor privacy terms as ministry documents, not fine print.
Frequently Asked Questions
Is using an AI chatbot for spiritual advice automatically wrong?
The Christianity Today argument is narrower than a ban. It holds that pastoral guidance implies accountability to a specific person, which a chatbot cannot supply — the technology may assist, but it should not be the shepherd.
What should a church do first if it already uses one of these tools?
Inventory what is deployed, identify the data each tool collects, and assign one named leader accountable for it. Then publish a short policy covering disclosure, crisis escalation, and what the tool must never handle.
Could a digital twin ever be done responsibly?
Possibly, under strict conditions: explicit consent from anyone whose words are used, clear labeling that the clone is not the pastor, no access to confidential disclosures, and a human follow-up path for anything serious.
Leave a Reply