Redact a PDF online - permanently black out names, IDs, account numbers and signatures

A rectangle, highlight or black text background in Word, a PDF editor or a notes app is a separate object laid over the page. The text underneath stays in the file: it can be selected, copied and found by search, and often it is enough to delete or move the rectangle to see everything. A black highlight in Word, saved as PDF, also leaves text that can be copied.

This tool works differently. It draws each page with marks as an image, burns the boxes into the pixels and builds a new PDF page from that image. In the result there is no text under a box, because a solid colour took its place.

• Pages with marks become images (JPEG or PNG, 150, 200 or 300 DPI). The page size stays the same, but the page loses its text layer: you cannot search it or select text on it, and links and form fields disappear.
• Pages without marks are copied unchanged, with their text, links and images at the original quality. Bookmarks and the document outline are not carried over to the new file.
• A password-protected file: its pages cannot be copied unchanged, so every page becomes an image and the result has no password.
• Metadata (title, author, subject, keywords) is removed by default. The name of the program that created the file is never kept. Turning the option off keeps the first four fields.

Redaction cannot be undone in the output file, so keep the original.

The search reads the document's text layer, the same one you use when you select text in a PDF viewer. The ready-made patterns recognize:

• PESEL (the Polish national ID number): 11 digits. If the checksum does not match, the match is flagged, because it may be a different number.
• bank account numbers: 26 digits, also with a PL prefix and with spaces or hyphens between digits,
• email addresses,
• phone numbers: mobile (9 digits, also with +48 or 0048) and landline with an area code.

Your own phrases, such as a surname or a street, are found regardless of letter case and accents. "Whole words only" skips matches inside longer words. Every match appears in a list and on the page preview. Untick the ones you do not want to hide and add the rest as boxes.

The search has limits. It cannot read text in a scan or a photo because it has no character recognition (OCR). Mark such a page by hand, or first add a text layer with OCR PDF. The position of letters inside a long piece of text is estimated, so look at every box in the preview and adjust it by hand when needed.

• The same data in other places: the header, the footer, next to a signature and on later pages. After saving, the tool checks the file for your searched patterns and phrases, but only in the text layer.
• Comments and other annotations on pages without marks stay in the file.
• The file name. The result gets a "-redacted" ending, but the rest of the name is unchanged, so "contract-john-smith.pdf" still reveals the name.
• Context. Hidden data can sometimes be guessed from what surrounds it: the length of a box, a date or a company name. Blacking out parts of a document does not guarantee that it is anonymous under data protection law.