Sanitize PDF
Large language models leave invisible characters behind: zero-width spaces, narrow no-break spaces, soft hyphens and bidirectional marks that survive every copy and paste. This removes them from a PDF in place, along with the metadata, comments, bookmarks and embedded attachments the file carries. The document stays a real PDF with selectable text, and the wording of your work is never altered. It is free, needs no account, and adds no watermark to your result.
Sanitize PDF
Add your files. Recommended settings are ready, so you can process in one click.
PDF, Word, images, audio, video, archives and more
Up to 100 files · 250 MB · No account- 1Add files
- 2Choose an action
- 3Download or keep going
What people use it for.
What happens to your file.
Every text object on every page is scanned for the invisible code points, and the affected spans are redrawn without them using the same font, size, colour and position, so the page looks identical. This edits the text objects themselves rather than rasterising the page, which is why the result stays selectable, searchable and accessible instead of becoming a picture of your document. Metadata, annotations, bookmarks and attachments are cleared in the same pass.
Where it stops: Visible characters are never touched: an em dash stays an em dash, because rewriting your punctuation would change your writing. Text baked into a scanned page image is not text objects, so run OCR first if your PDF is a scan.
Everything you need, without a paywall.
Common questions.
Will this change what my document actually says?
No. Only characters with no visible width are removed. Every word, every mark of punctuation and every line break you can see stays exactly as written, so the meaning and the reading of your work are untouched.
Why not just copy the text into a plain text box?
Because you lose the document. The tools that clean invisible characters almost all work on pasted text, so you get a wall of unformatted text back and have to rebuild headings, citations, tables and figures. This one edits the PDF itself and hands the PDF back.
Does the text stay selectable afterwards?
Yes. The cleaned spans are re-drawn as text with the original font embedded, not painted over or flattened to an image. The result is still searchable, still copyable and still readable by a screen reader.
Which characters are removed?
Zero-width space, zero-width joiner and non-joiner, the word joiner, narrow and ordinary no-break spaces, soft hyphens, bidirectional and directional formatting marks, and the byte-order mark. These are the code points that carry no width and are used to fingerprint machine-written text.
Is my file uploaded anywhere else?
No. Processing runs on our own server with no external service, no API key and no account, and the temporary working copy is removed after processing.
You might also need.
Remove common PDF document metadata fields on your device and save a cleaned copy without changing visible page content.
Permanently redact selected text or rectangular regions from PDF pages and burn the redaction into the saved output.
Turn scanned PDFs into searchable documents with OCR in over 100 languages. Pages are straightened and read at 300 DPI, the original scan stays untouched behind an invisible text layer, and pages that already hold real text are left exactly as they are.