Anonymising court judgments for publication: Find Case Law, family courts and open data

Open justice means judgments should be readable by anyone. It does not mean every child, victim, witness or bank account mentioned in them should be searchable forever. Anonymising court judgments is the craft of keeping the legal reasoning intact while removing what identifies the people involved. In England and Wales that job is shared between judges, court staff and The National Archives; for law firms, publishers, universities and public bodies that republish or reuse decisions, it becomes their own data protection responsibility.

This article explains how the official system works in the UK, which statutory protections make anonymisation mandatory in certain cases, how US federal courts approach redaction in filings, and what a sensible workflow looks like if you publish judgments yourself.

Find Case Law: who decides what is anonymised

Since 2022, Find Case Law, run by The National Archives with the Ministry of Justice, HM Courts and Tribunals Service and the Judicial Office, has been the official free source for judgments from courts and tribunals in England and Wales, plus UK-wide decisions of the Supreme Court and the Privy Council. Its privacy notice is clear about roles: the court or tribunal that decided the case is the controller for the content of the judgment, and The National Archives acts as processor for publication.

In practice this means anonymisation is a judicial decision. The judge decides whether names, addresses or other details are redacted, normally reflected in the rubric at the top of the judgment. The National Archives will not take down or alter a judgment without an instruction from the court. If you spot personal data that should not be there, the route is the court that issued the decision, not the archive.

When anonymity is required by law

Open justice is the default, but several statutes override it. The most relevant for anyone publishing judgments are:

• Sexual Offences (Amendment) Act 1992, section 1: complainants of sexual offences have lifelong anonymity; nothing likely to lead members of the public to identify them may be published.
• Children Act 1989, section 97 and Administration of Justice Act 1960, section 12: strict limits on publishing information about children involved in family proceedings and on material from private hearings.
• Court-specific anonymity orders and reporting restrictions, which bind anyone who republishes the judgment, not only the press.

For family cases, the President of the Family Division issued Publication of Judgments Practice Guidance in June 2024, replacing the 2014 and 2018 guidance. It sets out how to anonymise children judgments and financial remedies judgments, recommends a pseudonym or an initial that is not the person's real initial, and devotes a whole section to avoiding jigsaw identification: the way a child's sex, ethnicity, school, local area or a sibling group can identify a family when combined, even with no names left.

A note on US federal courts

In the United States the emphasis is on the filings rather than the judgment. Federal Rule of Civil Procedure 5.2 requires that electronic and paper filings show only the last four digits of a Social Security or taxpayer number and of financial account numbers, only the year of birth, and only the initials of a minor. Anyone building a dataset from court records should treat these as the minimum, not the maximum, of what to remove.

What to remove and what to keep

The family guidance makes a useful point: anonymisation is not a zero-sum game. Removing one non-essential fact, such as the town or the number of siblings, can avoid having to redact something that matters more for understanding the case. The concept behind all this is explained in our glossary entry on re-identification risk.

Consistency across the document

If the father is "F" in paragraph 12 and "Mr X" in paragraph 40, readers lose the thread and may infer identities by elimination. Every person should keep one stable alias throughout the judgment and across related decisions in the same case. That is exactly what consistent pseudonymisation with a saved mapping is designed to do.

Mistakes that leak data when you republish

• Drawing black boxes over a PDF without removing the text underneath, so it can be copied or extracted.
• Missing headers, footers, cover sheets and service details that carry names, addresses and full case numbers.
• Publishing scanned orders without OCR, so automated checks cannot see the text.
• Leaving document metadata (author, original file path, tracked changes) in the published file.
• Linking to related proceedings that, when searched, return an unredacted version.

Where software helps

Publishers, legal teams and courts that handle many decisions cannot rely on manual redaction alone. Anonimatum detects names, organisations, addresses, dates, amounts and vehicle registrations with its own AI, plus pattern-based identifiers such as emails, phone numbers and bank details, and lets you add custom patterns for local formats. A justice-oriented substitution profile replaces each person consistently, an allow list keeps terms that must stay (the court's name, the judge), and a preview lets a person review every change before it is applied. Text is processed on servers in the European Union and is not sent to external language-model APIs. See how it works for anonymising judgments and court orders.

The judgement call about what is material to the reasoning stays human. What the tool removes is the mechanical work and the risk that a bank account in an annex or a surname in a footer slips into the public version.