# Accuracy & honesty

A score is a reading, not a ruling. ScanForAI shows the bands plainly and says where it's wrong, because false accusations are the real harm.

Bands (low → high): **Reads human → Some AI patterns → Leans AI → Reads like AI.**

**No detector is 100% accurate — including this one.** Formal, technical, or non-native human writing can score high; edited or newer AI can score low. Treat the score as one signal, never as the sole basis for a decision about a person, and never to accuse.

## The honest failure modes

- **Formal & non-native writing.** Formal, academic, or non-native English writing uses fewer contractions and more even sentences, so it can read high. ScanForAI eases off those signals by register, but the risk never fully disappears.
- **Edited or newer AI.** A light paraphrase or edit removes most surface tells, and the newest human-tuned models can slip past almost entirely. A low score means the text *reads* human — not that it was.
- **Short text.** Under ~300 words most signals abstain or wobble. Give it a few solid paragraphs of prose for the most reliable read.

Never use a result as the sole basis for an academic, disciplinary, hiring, or other consequential decision about a person, and never to accuse someone.

Source: https://scanforai.com/accuracy.html