AI detection guide
What the scores mean, how to read the highlights, and when a result is worth trusting.
What the AI probability means
It is the estimated chance that the content was generated by an AI model. A high score is a flag, not proof; a low score is not a guarantee of human authorship.
How to read the signals
Perplexity measures predictability, burstiness measures sentence-to-sentence variation, and heuristics measure surface statistics. The fused score weighs them together.
When to trust a result
Trust the result most on longer texts and clear-cut cases. Short texts, edited AI output, and translations are the hardest cases — treat them as uncertain.
Using results fairly
Never use a single score as sole proof of misconduct. Treat detection as a screening step and confirm with a human conversation or further evidence.