Free browser-based analysis suite
Understand the text behind an AI detector score.
Inspect writing patterns, compare conflicting detector results, and prepare clean text from documents without uploading the writing.
Local linguistic analysis
See the patterns behind the writing.
Measure rhythm, repetition, vocabulary, and structure without turning uncertain signals into a fake authorship verdict.
Waiting for text
Review next
Compare detector results
Understand disagreement, not just averages.
Enter results from two or more detectors. The explainer shows spread, agreement, and a responsible next step.
Detector scores are probabilistic and should not be treated as proof.
Private file extraction
Prepare a clean sample from a document.
Extract text from PDF, DOCX, TXT, MD, and CSV files, then remove common document noise before analysis.
A better workflow
Use detector output as evidence to inspect, not a verdict to repeat.
AI detection is sensitive to the text sample, document cleanup, language, and detector choice. The lab keeps those uncertainties visible instead of hiding them behind a single confident number.
- Prepare the sampleRemove repeated headers, page numbers, and optional references.
- Inspect the writingLook at rhythm, repetition, vocabulary, and structural regularity.
- Compare the scoresPreserve disagreement instead of treating an average as certainty.
Built for mobile work
Move from a desktop draft to your phone.
Use the web lab for quick private analysis, then continue in the AI Detector app for camera capture, document scanning, and saved checks.
Questions and limits
What the AI Text Lab does—and does not—claim.
Does this tool tell me whether text was written by AI?
No. The writing pattern analyzer describes measurable features such as sentence variation, repetition, vocabulary, and transitions. Those signals can appear in both human and AI-assisted writing and are not proof of authorship.
Is my text uploaded or stored?
No. Writing analysis and document extraction run in your browser. The site records anonymous tool actions such as starting or completing a module, but it does not send your text, file contents, scores, or report to the server.
Why do different AI detectors give different scores?
Detectors use different models, thresholds, training data, and text preparation rules. Short samples, quoted material, references, highly formal prose, and non-native English writing can also change results.
Can the document tool read scanned PDFs?
The current version extracts selectable text from PDF and DOCX files. Image-only scans need OCR and are not silently treated as empty text. The mobile app is the better workflow for photos and scanned pages.
Should I average several detector scores?
An average can summarize the numbers, but it can hide substantial disagreement. The score explainer shows both the average and the full range so you can see how sensitive the result is to the detector used.