AI Detector Open the detector

Import locally

DOCX and PDF AI detection without sending the document to a detection server

AI Detector can read supported document formats in the current browser before analyzing their extracted text. The import path is designed for drafts and files that should not be posted to a remote detection API.

Supported file formats

  • TXT and Markdown files.
  • DOCX Word documents.
  • PDF files that contain a readable text layer.

You can also paste text directly. The current product accepts up to 15 MB per imported file, up to 200 PDF pages, and up to 10,000 extracted characters per check.

What happens after you choose a file?

  1. The browser reads the selected file locally.
  2. The importer extracts readable text and checks the product limits.
  3. The supported text is routed for language-specific analysis.
  4. The result shows signals, coverage, local passages, and text metrics.
Privacy boundary

The source document is not posted to a detection server. The site still needs network access to load the website and the resources used by the browser analysis. Local history summaries remain in this browser and can be cleared.

Why a scanned PDF may not work

Many scanned PDFs contain images of pages rather than a text layer. Without readable text, the importer cannot provide reliable input to the detector. Run OCR using a tool you trust first, review the extracted text, and then decide whether it is appropriate to analyze.

Extraction can also change line breaks, whitespace, or reading order. Review the imported text before treating a result as representative of the original layout.

File analysis is still not authorship proof

Importing a DOCX or PDF does not change the detector’s limitations. Short, translated, formula-heavy, code-like, or heavily edited text can produce false positives or false negatives. The result cannot prove who wrote the document, which product created it, plagiarism, or academic misconduct.