How to Redact a Subject Access Request
A subject access request means disclosing a person's data without disclosing everyone else's. What UK and EU GDPR ask you to remove, what you must leave in, and how to redact the bundle so it stays redacted.
In short
Under Article 12(3) of the UK and EU GDPR a subject access request must be answered within one month, extendable by two further months for complex or numerous requests. Article 15(4) says the copy must not adversely affect the rights and freedoms of others, and in the UK, Schedule 2 paragraph 16 of the Data Protection Act 2018 lets you withhold another person's data unless they consent or it is reasonable to disclose without consent. The ICO says exemptions cannot be applied routinely or in a blanket fashion, so the normal answer is to redact the third-party data and disclose the rest, not to withhold whole documents. Redaction must remove the content, not cover it, and the requester's own data must stay in.
A subject access request asks you for one person's data, and almost every document that holds it also holds someone else's. The email about a grievance names the colleagues involved. The case note about a child names the parents, the social worker and the neighbour who called. Answering the request means disclosing the first without disclosing the second, and doing it to a deadline. That is a redaction job, and it is where most of the time in a subject access request goes.
One month, and what changed in 2025
The deadline comes from Article 12(3) of the GDPR, which the UK kept in the UK GDPR. You must respond without undue delay and in any event within one month of receiving the request. That can be extended by two further months where necessary, taking into account the complexity and number of requests, but you must tell the requester about the extension, with reasons, inside the first month. A large request with hundreds of documents to redact is the ordinary case for that extension, not an excuse for missing it.
The UK has since adjusted the process. The Data (Use and Access) Act 2025 makes explicit that organisations need to make reasonable and proportionate searches rather than exhaustive ones, and adds a stop-the-clock rule that pauses the response time while you wait for information you need from the requester. The government has said these changes commence in stages through regulations, so check the ICO's current guidance for the position on the day you respond. None of it changes the redaction work. Whatever you find, you still have to clean it before it goes out.
The rule: other people's rights come with the copy
The rule that drives the redaction is Article 15(4): the right to obtain a copy shall not adversely affect the rights and freedoms of others. In the UK this is made concrete by Schedule 2, Part 3, paragraph 16 of the Data Protection Act 2018. You do not have to disclose information that identifies another individual unless that person has consented, or it is reasonable in all the circumstances to disclose it without their consent.
What counts as reasonable is a judgement with listed factors: the type of information, any duty of confidentiality owed to the other person, the steps you took to seek their consent, whether they are capable of giving it, and whether they have expressly refused. A colleague acting in their professional role, named in a routine email, is often reasonable to disclose. A witness in a complaint, a family member in a social care file or a patient in a clinical record usually is not. That decision belongs to a person who knows the context. No tool should make it for you.
Redact the person, disclose the document
Note what paragraph 16 does not say. It is not permission to hold back a document because it mentions someone else. The ICO is explicit that you cannot routinely rely on exemptions or apply them in a blanket fashion, and its guidance on finding and supplying information says to redact third-party information where necessary and to disclose only part of an email if that is what it takes. Redact the other person, disclose the rest. Withholding a whole email to protect one name in it is the pattern that turns into a complaint.
The requester's own data stays in. It sounds obvious and it is the most common over-redaction in practice. A requester who receives their own name, address and case reference blacked out has received nothing, and has grounds to complain that the request was not answered. Any automatic detection will flag the requester as readily as everyone else, because to a detector a name is a name, so the review step has to put them back.
Where third-party data hides
Third-party data hides in places a reader skims past. Email headers carry names and addresses in the From, To and Cc lines, and signatures carry job titles, direct lines and mobile numbers. Forwarded chains repeat all of it several times. Spreadsheets hold other people's rows next to the requester's. Scanned letters carry names in handwriting and in letterheads. Attachments carry their own metadata, including the author and the person who last edited them. A redaction pass that covers the body text and misses these has not finished.
Remove it, do not cover it
Once you know what to remove, the redaction has to actually remove it. A black box drawn over text in a PDF, a black highlight in Word or a cell coloured black in Excel leaves the text in the file, readable by anyone who selects it, copies it or opens the file in another program. In a subject access request the reader is the one person most motivated to look underneath, and the information underneath is, by definition, about someone who did not agree to its disclosure. We covered how these failures are recovered in how to unredact a PDF, and how to do it properly in how to permanently redact a PDF.
Keep a redaction log
Keep a record of what you redacted and why. If the requester complains to the ICO, the question will be why each redaction was made, and the answer needs to exist before anyone asks. A log that pairs each removed item with the exemption relied on, usually paragraph 16 for third-party data, is the cheapest defence you will ever write. Keep the unredacted originals unchanged and write the redacted versions as separate files, so the record and the disclosure can never be confused.
Where the redaction runs
The documents in a subject access request are personal data of the most sensitive kind an organisation holds about an individual, gathered precisely because they concern that person. Uploading the bundle to a cloud redaction service makes that service a processor, which needs a contract under Article 28 and, for a public body, often a data protection impact assessment before it can be used at all. For many teams, redaction that runs on the machine where the documents already sit is the simplest way to avoid adding a processor to every request.
That is what Omit Redact does. It finds names, addresses, phone numbers, email addresses, organisations and structured identifiers in PDFs, Word documents, spreadsheets, text and scanned images, with detection that runs entirely on your Windows machine and no network access at runtime. You review every detection before anything is written, so the requester's own details are kept and the reasonableness calls stay with you, and the output is a separate file with the content removed rather than covered. Its page for public sector redaction covers the wider records work, and HR and consulting covers employee requests, which are the other place most subject access requests land.