# Is Genesis 1 AI-generated? 7 of 7 AI detectors said yes

Source: https://toolbenchmarks.com/famous-texts/genesis-1/  
Tested: 30 October 2026. Publisher: Tool Benchmarks. Data licence: CC BY 4.0.

> No. The text we tested is Genesis 1 from the King James Bible, published in 1611, centuries before AI existed. All 7 of the AI detectors we tested still flagged it as AI-generated, the only one of our 30 famous texts that every detector flagged. Desklib, the detector with the fewest false positives among those that catch a real share of AI text, gave it 99.8%.

Genesis 1 (King James Version), King James Bible translators, 1611. Flagged as AI by 7 of 7 detectors.

| Detector | Verdict |
|---|---|
| Fast-DetectGPT (Qwen2.5-1.5B) | AI-generated |
| Desklib AI text detector | AI-generated |
| Fakespot AI text detector | AI-generated |
| MAGE | AI-generated |
| E5-small LoRA detector | AI-generated |
| OpenAI RoBERTa detector | AI-generated |
| TMR (Target Mining RoBERTa) | AI-generated |

Passage tested: "In the beginning God created the heaven and the earth. 1:2 And the earth was without form, and void; and darkness was upon the face of the deep. And the Spirit of God moved upon the face of the waters. 1:3 And God said, Let there be light: and there was light. 1:4 And God saw the light, that it..."

## Every detector called it AI

Genesis 1 is the only passage among our [30 famous texts](/famous-texts/) that all seven detectors labelled AI-generated. The classifiers that report a probability were close to certain. Desklib put it at 99.8%, TMR at 98.5% and Fakespot at 98.2%, and the OpenAI RoBERTa detector's score rounds to 100.0%. The E5-small LoRA detector was the least sure of them at 82.9%. MAGE and Fast-DetectGPT flagged it as well.

Desklib flagged 6.3% of the human texts in our main test and the OpenAI RoBERTa detector 4.8%, the two lowest rates we measured, and the RoBERTa detector flagged no other famous text at all. TMR, by contrast, flags almost anything with paragraph breaks. On Genesis 1 the most cautious detectors and the least cautious ones gave the same answer, and it was wrong.

## Why the King James Genesis looks machine-made

We can only offer likely explanations, but this passage has several features detectors seem to react to. It is built from a handful of formulas that come back for every day of creation: "And God said, Let there be", "and it was so", "and God saw that it was good", "And the evening and the morning were". Once a reader has seen the first day, much of each later day is easy to predict word by word, and Fast-DetectGPT reads low surprise as a sign of AI.

The King James Bible is also one of the most reprinted texts in English, so the language models behind these detectors have probably met this chapter many times in training. A model that has seen a passage often finds it predictable, and predictable text looks machine-made to these tools. The vocabulary is plain and the sentences mostly start with "And". The copy we tested keeps the verse numbers and puts each verse on its own line, which is the multi-paragraph layout that TMR flags almost every time.

## Other writing at the same risk

Writing that repeats set phrases or quotes very familiar sources is probably at the same risk. The [Declaration of Independence](/famous-texts/declaration-of-independence/), another text nearly everyone has read, drew four flags. Our [false positives page](/false-positives/) shows how often these detectors flag ordinary human writing too.

## Questions people ask

### Why do AI detectors say the Bible is AI-generated?

The likeliest reasons are repetition and familiarity. Genesis 1 repeats the same few formulas for each day of creation, and the King James text is so widely copied that language models have very likely seen it many times. Both make the wording very predictable, which detectors tend to read as machine writing.

### If every detector agrees, is the verdict more reliable?

Not on this evidence. All seven detectors we tested agreed that Genesis 1 is AI-generated, and all seven were wrong. Agreement between tools shows only that the text has features they all react to.

### Did the OpenAI detector flag Genesis 1?

Yes. The OpenAI RoBERTa detector flagged it, and it flagged no other famous text we tested. It flagged 4.8% of human texts in our main test and catches little modern AI writing.