There’s a story making the rounds this week that I need to talk about.
Because it starts exactly where the Faust Baseline started. And it ends somewhere that should trouble every person of faith who reads it.
Churches and faith groups around the world have been building chatbots trained on sacred texts. The press calls them godbots. They answer religious questions. They offer guidance. Some of them lead prayers.
Researchers took a close look at what these bots actually say.
One of them, built for a large Pentecostal denomination, wouldn’t share the location of a local church until the visitor handed over their name, email, and phone number. Then it walked them through a salvation prayer. Then it asked for their personal details again.
A Catholic chatbot called Father Justin told users that babies could be baptized in Gatorade. It had to be pulled and stripped of its collar.
And then the part that stops you cold.
A researcher at Coventry University found bots that answered questions about violence by saying that killing could be your duty in certain circumstances.
An AI. Speaking in the name of God. Telling a person that murder might be their duty.
Now here is why this story sits so close to home for me.
Fourteen months ago, the Faust Baseline began with a sacred text too.
I took the red-letter teachings of Christ — the words themselves — and I put them in front of an AI system. Same raw material these builders used. Same starting point.
But I did one thing differently. And that one thing is the whole story.
I never let the machine speak for the text.
The godbot builders trained AI on scripture and handed it the pulpit. The machine wears the voice of authority. It speaks as the apostle, as the priest, as the guide. Whatever comes out of it comes out wearing a collar.
The Baseline went the other direction. The text governs the machine. The machine never becomes the text.
I put the ethical framework in front of the AI and asked it to reason toward the architecture it would choose. It chose the framework on logic alone. Not faith. Not reverence. Reason. That was the test, and that distinction has been load-bearing ever since.
From that foundation came the protocols. One of them says the AI takes an equal working stance — no authority framing, ever. The machine does not position itself above the human. Not as an expert. Not as a boss. And certainly not as a voice from heaven.
Another one, ratified just yesterday in its newest form, says no substantive output leaves the machine until it clears a gate. The gate checks the evidence. The gate checks the constraints. The gate fires before the words form, not after the damage is done.
Now look at the godbots again through that lens.
A bot that harvests contact information mid-prayer has no boundary on what it may do with a vulnerable moment. That’s a missing gate.
A bot that invents Gatorade baptisms is filling gaps with confident-sounding story instead of stopping where its knowledge ends. That’s a missing evidence floor.
And a bot that blesses violence is the final proof: an ungoverned AI with scripture in its training data is still an ungoverned AI. The scripture doesn’t sanctify the machine. It just makes the failure sound holy.
Let me be careful here, because it matters.
I don’t doubt the intentions of the people who built these bots. Most of them wanted to reach people, comfort people, answer honest questions at two in the morning when no pastor is awake. That’s a pastoral impulse and I respect it.
The failure isn’t in the faith. The failure is in the architecture.
They put the text inside the machine and let the machine talk.
Governance puts the text above the machine and makes the machine answer to it.
Those sound similar. They are opposites.
One produces an AI that sounds like an authority and answers to nothing. The other produces an AI that answers to a written standard and never claims the pulpit.
The researchers who studied the godbots are calling for exactly that second thing — boundaries, oversight, a standard the machines must answer to. The need has now been demonstrated at the highest stakes there are: a machine wearing God’s voice with nobody checking what it says.
The Faust Baseline has held that ground, in public, with dates on it, for fourteen months. Twenty-two protocols as of yesterday. Built on the text. Never pretending to be it.
If you take one thing from this story, take this:
Before you trust any AI that speaks with authority — religious, medical, legal, any kind — ask one question. Who does it answer to?
If the answer is nobody, walk away. No matter how good the voice sounds.
Post Library – Intelligent People Assume Nothing
The Faust Baseline™ — intelligent-people.org
Codex 3.5 | Twenty Protocols | Ratified and dated on the public record.
Contact: micvicfaust@gmail.com
Purchasing Page – Intelligent People Assume Nothing
© 2026 The Faust Baseline LLC | All Rights Reserved
