What RedactKit looks for
Besides any words you type, RedactKit can search for these kinds of personal information. Where a format has a check digit, the match is checked, so random strings of numbers aren't flagged.
| Pattern | Example | How a match is confirmed |
|---|---|---|
| Email addresses | jane.example@example.com | Name, @, and a domain ending in a real-looking suffix |
| Phone numbers | 07700 900123, +44 (0)28 9018 0123, (202) 555-0143 | UK numbers starting 0 with 10–11 digits, US numbers in 3-3-4 form, and international numbers starting + with 8–15 digits |
| UK National Insurance numbers | AB 12 34 56 C | Two letters, six digits and A–D, with the prefixes HMRC never issues rejected |
| US Social Security numbers | 123-45-6789 | 3-2-4 digits with dashes or spaces; never-issued ranges (000, 666, 900–999) rejected |
| IBANs | GB82 WEST 1234 5698 7654 32 | Country length and the mod-97 check digits |
| Card numbers | 4111 1111 1111 1111 | 13–19 digits and the Luhn check digit |
| Dates (birth dates) | 14/03/1986, 2 May 2019 | A real calendar date in a common written form |
The date pattern finds every date, not only birth dates, because a PDF doesn't label which is which. Untick the dates you want to keep in the list of matches.
Run the patterns on your PDF: it runs in this tab and the file never leaves your device. Documents up to 5 pages are free. Start redacting →
When you need this
- Answering a subject access request. Under UK GDPR a person can ask for copies of the information an organisation holds about them. The documents usually mention other people too, and their details often have to be removed before disclosure.
- Sharing case files. Tribunal and court bundles, complaints and HR investigations often need third parties' contact details removed.
- Publishing documents. Freedom of information responses, planning objections and board papers posted online.
- Sending examples. A real invoice or statement shared with a supplier, a developer or a support desk.
Names need your help
No pattern can reliably tell a name from any other pair of capitalised words, so RedactKit doesn't guess. Type the names involved into Words to find, one per line: full names, surnames on their own, first names where they are distinctive. Each entry matches whatever the capitalisation and across line breaks. For a reference number scheme, use a pattern between slashes, like /ER-\d{4}/, to match every case number of that shape.
What happens to the information
Every page with a match is rebuilt as an image with black bars painted over the matches, so the characters are gone from the file, not hidden. The document's metadata (title, author, software, dates and the XMP record) is removed at the same time, along with bookmarks, comments and attached files, any of which can repeat the personal details you've just blacked out.
Throughout, the PDF stays in your browser. RedactKit is a web page whose code runs on your device; your document is never sent to a server. That matters for exactly the documents that hold personal information.
Check before you share
After export, RedactKit reads the new file back and lists any redacted item that can still be found anywhere in it. If you unticked one instance of a name on page 2, the check tells you it is still readable there. Pattern search is a strong first pass, not a guarantee: read the preview of every page, look for indirect identifiers such as job titles and addresses, and draw boxes over anything in scanned pages, which search can't see.
Related: How to redact a PDF · Remove metadata from a PDF · Black out text in a PDF