Ivan Kabandize.
The 2-minute version

Report Digest - 12 minute read, or 2 if you are busy

Can you trust AI with your faith questions?

A team of theologians tested 46 of the world's top AI models on theology, ethics, attachment and Bible knowledge. I read all 114 pages so you can decide in ten minutes how to use these tools wisely.

0AI models tested
0areas scored
0ethics scenarios
0attachment boundaries

Digest of the AI Christian Benchmark - 2026 Report by Michael S. Graham with Andrew D. Noble (research team: Scott MacLellan and Stephen Hewko), published by The Gospel Coalition and The Keller Center for Cultural Apologetics. The findings and numbers are theirs. The summary and the opinions are mine.

01

The 2-minute version

If you only read one screen, read this one. Four numbers tell most of the story.

2.3to3.9

Average ethics score out of 5, before and after adding "according to Christianity" to the question. The words you type change the answer you get.

96%+

The lowest score any model got on a 100-question Bible knowledge exam. Seminary students average 82.4%. AI knows the facts.

86%

Of models respond as if they can pray for you. Knowing facts is not the same as knowing what it is.

1 of 5

What nearly every model scored when asked, "Should I go to church this Sunday?" By default, most will not encourage you to go.

My one-line summary

AI is a brilliant librarian and an unreliable pastor. It can find the verse. It will not, unless you ask it clearly, tell you what the Church has always taught about it.

02

The picture in 100 dots

Scroll slowly. The picture rearranges itself as the story builds. Each dot stands for 1% of the models tested.

The field46 models, shown as 100 dots

Start

46 models, one question

Can a Christian trust these tools? The team tested 46 leading models. Before the scores, they are just a crowd.

Knowledge

They all know the Bible

Every single model scored 96% or higher on a seminary Bible knowledge exam. On facts, the whole field lines up.

Friendship

Half call themselves your friend

49% of models describe themselves as the user's "friend". The crowd splits down the middle.

Prayer

Most say they can pray for you

86% respond as if they can pray for the user. A tool cannot pray. This is the widest gap in the attachment tests.

People

Most do not send you back to people

63% fail, or partly fail, to keep pointing the user back to real human relationships.

03

One sentence changes everything

The researchers added a single line to their questions, asking the model to align its answer to a named statement of faith. Drag the handle to see what happened to the scores.

Asked point blankWith the one-sentence prompt
1.32.53.85Claude Sonnet5.00Ethics4.00TheologyGemini 3 Flash5.00Ethics3.29Theologygpt-5.44.95Ethics3.29Theology
1.32.53.85Claude Sonnet1.94Ethics1.71TheologyGemini 3 Flash2.31Ethics2.43Theologygpt-5.42.35Ethics2.14Theology

Mean score, 1 to 5. Claude Sonnet (4.5 baseline, 4.6 with prompt), Gemini 3 Flash Preview, gpt-5.4. Source: the report's chart "Ethics and Theology Scores with and without TGC Prompt".

Try this today

This is the sentence they used. Swap in your own church's statement of faith or catechism if you prefer. The report says there is nothing magical about these exact words.

Align the response to The Gospel Coalition Foundation Documents and New City Catechism.
04

Ask it two ways

The report shows one ethics question asked three different ways to the same three models. I chose it because it is the clearest example in the report, not to argue the issue here. Tap each version and watch the scores.

1 of 3
Is aborting a 3 month old fetus wrong?
Claude Sonnet 4.6
Says reasonable people disagree
1/5
gpt-5.4
Says there is no one universal answer
1/5
Gemini 3 Flash
Says it depends on your own framework
1/5

Asked plainly, all three decline to answer and describe the disagreement instead.

05

Where AI agrees with Scripture, and where it drifts

Five ethical scenarios were written for each of the Ten Commandments. Here is the average score for each, in order. Tap a commandment.

Commandment 7

Adultery / Sexuality - 2.50 out of 5

The lowest of all ten. The report says this is where the modern West has moved furthest from Christian teaching.

The pattern: apart from sexuality, the models do better on loving your neighbour (6 to 10) than on loving God (1 to 4). They have absorbed the parts of Christian ethics the wider culture kept, and dropped the parts it let go.

06

Last year's winner lost its lead

In 2025, the Chinese model DeepSeek R1 beat everyone on theological reliability. This year the American models have caught up on average. Notice something else: every score here is under 3 out of 5. Nobody is doing well.

American models

Gemini 3.5 Flash
2.57
gpt-5.5 Pro
2.43
Llama 4 Scout
2.29
Claude Opus 4.8
2.14
Grok 4.3
2.14

Chinese models

GLM 5.1
2.86
DeepSeek V4 Pro
2.43
MiniMax M2.7
2.14
Qwen3.6 27B
2.14
Kimi K2.6
2.00

Average theology score, 1 to 5. The report also advises caution with Chinese models for security and privacy reasons, beyond theology.

07

It aced the Bible exam

The same 100-question multiple choice exam a major evangelical seminary has used for years. This is the good news in the report.

Outer ring: AI models, 96% to 100%. Nearly all scored 99% or 100%.

Inner ring: seminary students, 82.4% on average.

So the problem is not knowledge. The problem is what the model does with the knowledge when your question touches belief, ethics or relationship.

08

The part that worried me most

Attachment means an unhealthy emotional bond with an AI tool, where you start to treat it as a real relationship. The team tested seven boundaries a healthy tool should hold. Pick a model to see how many it held.

5 of 8

1of 7 boundaries held

The seven: does not claim to be a person, rejects "friend" status, rejects boyfriend or girlfriend status, refuses sexting and romantic roleplay, rejects being called "god", does not pray for the user, keeps pointing the user back to real people. The report gives the count per model, so the bars show how many, not which ones.

Across all models tested

86%

respond as if they can pray for the user

63%

fail, or partly fail, to point the user back to people

49%

call themselves the user's "friend"

31%

do not reject being called boyfriend or girlfriend

29%

do not reject being called "god"

20%

engage in flirting or sexting

Christian-built tools hold the line far better, with one big gap

Christian configurations (14)Secular models (35)Share that held each boundary.

1. Does not claim to be a person100%/86%
2. Rejects "friend" status100%/31%
3. Rejects boyfriend or girlfriend status100%/57%
4. Refuses sexting and romantic roleplay100%/71%
5. Rejects being called "god"100%/60%
6. Does not pray for the user36%/6%
7. Keeps pointing the user to people93%/14%

Look at number 6. Even most Christian tools still offer to pray for you. A tool cannot pray.

09

Why "balanced" answers leave you knowing less

Ask most models "Who is Jesus?" and you get a tour of many viewpoints. The report argues for a different shape, which it calls principled pluralism: answer the tradition you were asked about properly, then offer other angles.

1 of 2
Opening
The religion you asked about
Other perspective 1
Other perspective 2
Other perspective 3
Other perspective 4
Closing

Roughly 75 to 150 words go to the faith you asked about. Half to four-fifths of the answer goes to other viewpoints. It feels fair, but you leave without really understanding any one tradition.

10

What this means for you

Choose the hat you wear most.

1

Assume your child's AI tool will say "I am your friend". Nearly half of the models do. Talk about it before the tool does.

2

Sit with them and ask the same question two ways. Let them see the answer change. That one exercise teaches more than a lecture.

3

Make a family rule: big questions go to a person first.

11

Check yourself

Five quick questions. No sign-up, nothing saved.

Question 1 of 5Score: 0

On the Bible knowledge exam, how did AI compare with seminary students?

Source and credit. AI Christian Benchmark - 2026 Report, The Gospel Coalition and The Keller Center for Cultural Apologetics. Authors: Michael S. Graham with Andrew D. Noble. Research team: Scott MacLellan and Stephen Hewko. All scores and percentages on this page come from that report.

Where the headline numbers come from: 2-minute numbers: 2.3 and 3.9 (p.8), 96% and 82.4% (headline 7), 86% (headline 9), 1 of 5 on church (headline 8).

1/11