Yes. A detection score is a probability, so any detector can read a text wrongly. ScribeLens shows the sentences behind the score and the reason for each flag, so you can check the result yourself. This page explains what a score measures and how to read it fairly.
A score measures how closely the writing matches patterns common in AI-generated text: word choice, sentence rhythm, predictable structure, and how much the style varies across the document.
So a high score means "this text shares measurable patterns with AI-generated writing." It is a probability, not a statement about who wrote the text.
ScribeLens detects these patterns sentence by sentence. Every sentence is marked human-like, mixed, or AI-like, and each flag names its reason, so you can see exactly what the score is based on.
Some honest writing shares patterns with AI text. When you see a flag on writing like this, open the sentence view and read the reasons before you form a view.
Detector v4.0 was retrained on real student essays and on essays by non-native English writers. In our own testing it is markedly better at recognising such writing as human. Reference lists and citations are set aside automatically, so they do not affect the verdict.
A low score means the text did not show strong AI-like patterns in this scan. AI text that has been heavily rewritten or run through a paraphrasing tool is harder to detect.
So read a low score the same way as a high one: as evidence to weigh together with what you know about the writer.
Read the evidence first and the number second.
See which sentences were flagged and which pattern each one shows. Compare them with the author's other work and drafting history. Then talk to the author. A detector shows you where to look; the decision stays with you.
AI writing leaves measurable patterns: uniform sentence rhythm, predictable word choice, formulaic structure. ScribeLens detects those patterns and shows where they appear, sentence by sentence, with a reason for every flag.
The headline verdict always agrees with the percentages printed beside it. Every report includes a note headed "How to use this result", and the exported PDF has a reading guide, so a teacher or supervisor can read the report without an account.
ScribeLens detects the measurable patterns AI writing leaves and shows them sentence by sentence. We do not publish a single accuracy figure, because results vary with text length, subject, and writing style. Each result is a probability, and the sentence-level evidence lets you check it.
No. A result shows that text shares patterns with AI-generated writing. It is evidence for a person to weigh together with drafting history and a conversation with the author.
Detector v4.0 was retrained on real student essays and on essays by non-native English writers. In our own testing it is markedly better at recognising such writing as human. Every flag still shows its sentence and its reason, so anyone can check it. ScribeLens analyses English text only.
Show your process: drafts, version history, notes, and outlines. Ask the reviewer to look at the flagged sentences and their reasons, not only the overall score.
Run a text and inspect the sentence-level signals behind the verdict. Without an account a scan covers 2,000 words; a free account scans a whole document.