TL;DR: Every major church management software vendor has shipped AI features into their product in the last twelve months, and adoption is moving faster than governance. Sixty percent of church leaders personally use AI monthly, but only thirty-three percent of churches use it in ministry operations, and only five percent have a written AI policy in place. The decision facing every church right now is not whether AI is coming into the church management software they already run, it is already inside it, but what to ask the vendor before turning the features on.
What You Need to Know About AI Inside Church Management Software:
- Planning Center connected its People product to ChatGPT and Claude via MCP in March 2026, then went live in the ChatGPT App Store in May 2026, making one-click connection of church member data into a foundation model the default friction state.
- Pushpay launched AI for People Search to general availability in April 2026, alongside AI for Giving Data, Studio AI, and the customer-support assistant Genny, with an AI involvement summary on the 2026 roadmap.
- Subsplash shipped Trends AI and the AI People Assistant in March 2026, layering natural-language search and consolidated analytics across the platform.
- Eighty-three percent of church leaders cite data privacy as their primary AI concern, sixty-four percent say a written AI policy is important, and only five percent of churches have one (Barna and Pushpay, 2026 State of Church Tech).
- The questions to ask the vendor are not “is this safe,” they are “where does our data go, who can it train, what permissions carry through, and how do we shut it off.”
What Changed Inside Your Church Management Software in the Last Twelve Months
Every major church management software vendor shipped meaningful AI features into their product in the last twelve months. Planning Center went live with ChatGPT and Claude integrations in March 2026 and listed in the ChatGPT App Store in May. Pushpay rolled AI People Search to general availability in April, and Subsplash shipped Trends AI and the AI People Assistant in March.
Planning Center’s connection runs through the Model Context Protocol, which lets staff use the People product directly from inside ChatGPT or Claude Desktop. The May ChatGPT App Store listing took setup down to one click, and the vendor’s roadmap signals MCP support for additional products beyond People.
Pushpay’s rollout is broader in scope. AI People Search reached general availability after a beta that logged 15,185 prompts across 331 accounts and 1,226 unique users. Alongside it, the company launched AI for Giving Data, Studio AI from Resi for sermon-to-clip generation, and Genny for customer support, with an AI involvement summary on the 2026 roadmap that aims to compress a five-minute profile review into a forty-five-second snapshot.
Subsplash took a third path. Trends AI is an analytics layer that consolidates data across the Subsplash platform and external sources with AI-assisted analysis, and the AI People Assistant adds natural-language member search. Pulpit AI handles sermon repackaging on the content side.
Tithely Church Management (the platform renamed from Breeze on November 11, 2025) and ChMeetings have not announced major standalone AI features in 2026. That is a posture, not a failure. Both platforms lean on simplicity rather than feature breadth.
Key Point: If the church runs Planning Center, Pushpay, or Subsplash, AI features are already inside the product, whether the staff has turned them on or not.
Why Most Churches Have Not Caught Up to Their Own Software
Adoption of AI by leaders is racing ahead of governance inside the churches they lead. Sixty percent of church leaders personally use AI at least monthly, but only thirty-three percent say their church is using AI in ministry operations. Sixty-four percent of leaders say a written AI policy is important, and only five percent of churches have one (Barna and Pushpay, 2026 State of Church Technology, released March 9, 2026).
The same survey of 1,306 U.S. church leaders found that only nineteen percent feel confident their church’s data is ready to produce meaningful AI insights, and that data privacy is the number one fear surrounding AI adoption.
ChurchTechToday put the consequence plainly in a piece published June 8, 2026: “If your church doesn’t have clear guidelines about how AI can and cannot use your members’ data, you’re effectively delegating that decision to whoever built the software you’re running.”
It is worth noting that the Barna and Pushpay study was funded by a vendor that sells AI tools to churches, which an industry watchdog has flagged. The methodology is sound, the conflict of interest is real, and both can be true at once.
Key Point: A church without an AI policy has not opted out of AI, it has handed the policy decisions to whichever vendors built its software.
What the AI Inside the ChMS Actually Touches
The AI features now embedded in the major ChMS platforms touch the most sensitive data a church holds: member names and contact details, attendance patterns, group memberships, giving history, children and youth records, and pastoral notes where staff have entered them. The data does not necessarily leave the ChMS, but it is being read, summarized, and queried by AI systems in ways that did not happen twelve months ago.
The Pushpay AI People Search beta surfaced what those queries actually look like. Across the 15,185 prompts logged during beta, six categories emerged: names and groups, giving and financial data, children and youth, group membership, attendance, and volunteering. Each carries a different sensitivity weight. A question about volunteer availability is not the same kind of data as a question about giving patterns, but in this generation of tools, both flow through the same AI surface.
Subsplash’s Trends AI consolidates data “across the platform and external sources”, a phrase worth reading slowly. The “external sources” part means the consolidation is not just the church’s data in isolation, it is the church’s data combined with whatever else the AI is pulling in to produce its insights.
Planning Center takes a different posture. The vendor states that permission boundaries carry through, so a staff member only sees what their Planning Center permissions allow them to see. A volunteer with limited access to People will not suddenly gain access to giving history because they connected the AI.
What the AI does not touch by default is what was never recorded in the ChMS in the first place. Pastoral counseling notes kept on paper, conversations on third-party platforms, and informal text threads sit outside the perimeter. The privacy risk lives where data has already been entered, not somewhere new.
Key Point: The AI inside the ChMS reads what the staff entered, so the privacy risk lives where the data has already been recorded, not somewhere new.
The Question Almost No One Is Asking, Where Does the Data Actually Go
When a church staff member connects Planning Center to ChatGPT through the App Store and starts asking questions about members, the conversation by default becomes training data for OpenAI’s foundation model. Planning Center says so itself in its own changelog, and the same applies to Claude through Anthropic. The default setting has to be changed inside the AI tool, not inside Planning Center.
The vendor’s own words are worth quoting directly: “By default, both Anthropic and OpenAI will train their models based on your conversations with them unless you change settings in your AI tool. In this case, your conversations could include personal and private information from your Planning Center account, and we strongly recommend that you don’t let these tools train their models on this data.”
In concrete terms, a user types “summarize the giving patterns of our top twenty donors,” and that prompt, along with the response, can become part of a training corpus that helps shape future model behavior. The setting that prevents it does not live in Planning Center, it lives in ChatGPT or Claude. A staff member who connects the integration without changing that setting is opting in to training by default.
The Brentwood Statement on AI and Christian Ministry, released June 8, 2026 by Lifeway and Southern Baptist Convention leaders, addresses this directly. The statement counsels pastors to be “cautious about feeding information into AI models that compromise the personal safety and security of congregations.” That is denominational guidance four days old at the time of this writing.
Pushpay’s posture is different. The company states that member data processed through Pushpay stays within its secure partner ecosystem and is not used to train external foundation models. That is the vendor’s statement of how the data is handled, framed accordingly. Independent verification of vendor data-handling claims is generally not available in this category, so the right reading is to take vendors at their word in writing and to keep asking until the writing is clear.
Key Point: The default setting on the consumer AI tool decides whether church member information ends up in a foundation model, and that default is opt-in to training.
The Five Questions to Ask the Vendor Before Turning AI Features On
The questions worth asking a ChMS vendor are not about safety in the abstract, they are specific, answerable, and tell the church what its data is actually doing. Five questions cover the ground that matters: data training, permission inheritance, deletion, audit, and shutdown. They sit inside the broader framework for evaluating any ministry technology vendor, but they are the AI-specific questions worth asking on top of that framework.
Where does our data go when the AI processes it, and is it ever used to train any model?
The right answer is a written one, not a sales-team verbal. Planning Center’s changelog disclosure is a model of the kind of clarity to ask for. If the vendor cannot point at a written policy of equal specificity, that is information.
Do the permissions in the ChMS carry through to the AI features?
When a volunteer with limited access uses AI People Search, do they still see only what their role allows? The right answer is yes, and the vendor should be able to point at where that enforcement lives in the product.
If we delete a member’s record, does the deletion propagate to anything the AI has indexed, summarized, or cached?
This question separates vendors who built AI on top of their data layer from vendors who built a parallel index that drifts out of sync. The answer matters for compliance, for pastoral care, and for the family that asked for their information to be removed.
What audit trail do we get when AI is used against our data?
Who queried what, when, and what did the AI return. Without an audit log, there is no accountability for an AI prompt the way there would be for a database query. A platform that cannot show its work is a platform that cannot be reviewed.
How do we shut the AI features off, today, without disrupting the rest of the platform?
The answer to this question reveals whether AI is a feature, with a clean off switch, or whether it has been wired into the core of the product. Both can be defensible architectures, but the church should know which one it is buying.
Key Point: A vendor that cannot answer these five questions in writing is not a vendor a church should hand its member roster to.
What Independent Voices Are Saying That the Vendors Are Not
The vendor announcements are loud and the independent voices are quieter, but the independent voices are where the most useful guidance lives right now. Three voices in particular are worth reading: the Brentwood Statement from Lifeway and SBC leaders, Carey Nieuwhof’s recent reframe of the AI conversation, and Lifeway Research’s own data on pastor skepticism.
The Brentwood Statement on AI and Christian Ministry articulates seven principles for ministry AI use: pursuing wisdom, championing God-given dignity, promoting truth and fidelity, cultivating trust and integrity, protecting privacy, representing and proclaiming the kingdom of God, and modeling faithful leadership. The privacy principle is the one most directly relevant to AI inside the ChMS, and the statement’s specific counsel against feeding congregational information into AI models is the clearest denominational guidance issued so far in 2026.
Carey Nieuwhof reframed the whole conversation in a piece for RELEVANT published June 11, 2026, one day before this writing. “I think we’re having the wrong conversation about artificial intelligence. Most of us are talking about what agent do you use? What prompts do you use? The bigger question is about the impact of artificial intelligence.” That reframe is worth carrying into every vendor conversation.
Lifeway Research adds the audience data. Eighty-four percent of pastors worry AI-generated content contains errors, eighty-one percent believe AI tools rely on unreliable sources, seventy-six percent worry about bias, and sixty-one percent of churchgoers are concerned about AI’s influence on Christianity. That skepticism is not noise, it is the audience response the technology is asking the church to address.
Key Point: The clearest counsel on AI inside the church is coming from voices that are not selling AI features.
What Is Still Being Figured Out, and What Is Already Real
Some questions about AI inside church management software are genuinely unsettled, and others have already been answered by events. The unsettled questions deserve patience. The answered ones deserve immediate attention.
The unsettled questions include whether ChMS-embedded AI counts as “data leaving the church” in a regulatory sense, whether industry-funded research like the Barna and Pushpay report should be weighted differently than fully independent research, and what an AI policy template looks like that does not become obsolete in twelve months. These are live debates with serious people on each side.
The answered questions are different. The NIST National Vulnerability Database documents a critical pre-authentication remote code execution vulnerability in ChurchCRM filed in 2026, alongside related advisories for insecure direct object references and SQL injection in the same product family. Those are concrete proofs that church member data sitting in any ChMS is a real target.
The First Baptist Church of Hammond ransomware breach is the most concrete recent example. A 2025 attack exposed information for 5,217 people, including Social Security numbers, state IDs, addresses, contact information, healthcare identifiers, dates of birth, and health information. The point is not that any specific ChMS will be breached, the point is that church member data is a real target and adding more AI surface to it is a decision with consequences.
The AI question and the security question are joined at the hip. They live in what happens at the seams between systems, and the seams are where most failures occur.
Key Point: The privacy stakes are not theoretical, they are documented in vulnerability databases and breach disclosures from the last twelve months.
Frequently Asked Questions
Is our church management software using AI right now?
Likely yes if the church runs Planning Center, Pushpay or its ChurchStaq products, or Subsplash. Planning Center connected its People product to ChatGPT and Claude in March 2026. Pushpay launched AI People Search to general availability in April 2026 alongside several other AI tools. Subsplash shipped Trends AI in March 2026. The features are usually opt-in to use, but the infrastructure ships enabled.
Does ChatGPT see our members’ data when we connect Planning Center?
When a staff member uses the official Planning Center connection inside ChatGPT or Claude, the AI sees what the staff member’s Planning Center permissions allow them to see. Planning Center warns in its own changelog that by default, OpenAI and Anthropic will train their models on those conversations unless the setting is changed inside the AI tool itself.
Do we need an AI policy if our church is not actively using AI yet?
Yes. Sixty percent of leaders personally use AI monthly, which means staff and volunteers are already using AI tools on their own devices to do church work, with or without a policy. A policy clarifies what data can and cannot be fed into AI tools, who can use which features, and how the church will respond when something goes wrong.
Are church management software vendors training their AI models on our church data?
Each vendor states a different posture. Planning Center is explicit that the AI tools themselves (ChatGPT, Claude) will train on conversations by default unless the user opts out inside the AI tool. Pushpay states that member data stays within its secure ecosystem and is not used to train external foundation models, framed as a vendor statement rather than independently verified. Always ask in writing.
What is the Brentwood Statement on AI?
A statement on AI and Christian ministry released June 8, 2026 by Lifeway and Southern Baptist Convention leaders. It articulates seven principles including privacy, truth and fidelity, and protecting the personal safety and security of congregations. The statement specifically counsels pastors to be cautious about feeding congregational information into AI models.
How do we shut off the AI features in our ChMS if we are not ready?
Each vendor handles this differently. Planning Center’s MCP connections are made by the user inside ChatGPT or Claude, so disconnecting the apps in those tools shuts off the integration. Pushpay’s AI People Search and similar features can be disabled at the account level. Subsplash’s Trends AI is governed by per-product permissions. Ask the vendor for a written shutdown procedure before turning the features on.
Where should we start if our church does not have an AI policy?
Three steps. First, take inventory of which AI features are already inside the church’s existing software stack. Second, decide which categories of member data (giving records, attendance, children and youth, pastoral care notes) are out of bounds for AI processing without explicit consent. Third, write a short policy covering what staff and volunteers can use AI for and what they cannot.
Key Takeaways: AI Inside Church Management Software in 2026
- Every major church management software vendor has shipped AI features in the last twelve months. Planning Center, Pushpay, and Subsplash all rolled production AI capabilities between December 2025 and June 2026.
- The features touch the most sensitive data the church holds. Names, attendance, giving, children and youth records, and group membership are all in scope for the new AI tools.
- Leader adoption of AI has outpaced church governance by a wide margin. Sixty percent of leaders use AI monthly, only five percent of churches have a written AI policy.
- The data-training default lives inside the consumer AI tool, not the church management software. A church that connects Planning Center to ChatGPT without changing settings in ChatGPT is opting in to training data by default.
- Five questions cover the ground that matters with any vendor: where the data goes, whether permissions carry through, what happens on deletion, what audit trail exists, and how to shut the AI off.
- The most useful guidance is coming from voices that are not selling AI features, including the Brentwood Statement, Lifeway Research, and independent commentators like Carey Nieuwhof.
Sources and References
- Barna and Pushpay, Technology for Missional Impact: 2026 State of Church Tech, March 9, 2026.
- Barna, How Church Leaders Are Using AI (And Their Top Concerns), March 26, 2026.
- Planning Center Changelog, Connect Planning Center People to Claude or ChatGPT, March 12, 2026.
- Planning Center Changelog, Planning Center is now in the ChatGPT App Store, May 28, 2026.
- Pushpay, Pushpay Launches New AI Tools to Transform Church Ministry, December 10, 2025.
- Pushpay, AI for People Search in ChMS, April 8, 2026.
- Pushpay, Pushpay’s Staq Transformation, June 2, 2026.
- Subsplash, Introducing Trends AI, March 25, 2026.
- Lifeway and Christianity Today, Young, Educated, and Urban Pastors Are Most Likely to Use AI, April 21, 2026.
- Lifeway and SBC, The Brentwood Statement on AI and Christian Ministry, June 8, 2026.
- MinistryWatch, The New Church-Tech Divide is Missional, Not Digital, March 30, 2026.
- ChurchTechToday, AI Ethics And The Church’s Most Sensitive Data, June 8, 2026.
- RELEVANT, Carey Nieuwhof: The Church Is Having the Wrong AI Conversation, June 11, 2026.
- NIST National Vulnerability Database, CVE-2026-39337 (ChurchCRM pre-auth RCE), filed 2026.
- Comparitech, Cybercriminals give Indiana megachurch 7 days to pay $600K ransom after data breach, July 29, 2025.
Where We Go From Here
The AI inside the church management software is not the church’s AI, it is the vendor’s AI running on the church’s data. The next decision is not whether to use it, the next decision is what to ask before turning it on, and what to write down about how we will use it.
If your ministry is working through how to evaluate the AI features now appearing in your church management software, or how to write a first AI policy that does not become obsolete in twelve months, we would be glad to think it through with you. We offer no-pressure consultations where we listen first, then share what we have learned helping ministries navigate the same questions. Read our pillar guide to church management software in 2026 for the broader landscape, or schedule a consultation.
