RegulationsUS · EU · UK

Test data rules, regulation by regulation

Few laws say "test data". Most of them still cover it: personal data doesn't stop being personal when it is copied into dev, and documents count as much as databases. Here is what each one asks of data outside production.

Three questions every regulator asks
  1. 1Is real personal data outside production, and why?
  2. 2Is it protected as well as it is in production?
  3. 3Can you show what was done to it, run by run?

At a glance

What each regulation says about test and non-production data, and which documents it touches most.

RegulationApplies toWhat it says about test dataDocuments most affected
GDPR and UK GDPR
EU and UK
Anyone processing personal data of people in the EU or UKMinimise it (Art. 5(1)(c)), build in pseudonymisation (Art. 25), secure it (Art. 32). Pseudonymised data is still personal data (Recital 26).CVs, emails, contracts, KYC packsRead more
HIPAA
US
Covered entities and business associatesNo test-environment exemption. De-identify by Safe Harbor or Expert Determination (45 CFR 164.514).Lab reports, discharge summaries, EOBs, claimsRead more
PCI DSS v4.0.1
Global
Anyone storing, processing or transmitting card dataNo live PANs in pre-production unless protected as the CDE (6.5.5). Remove test data before go-live (6.5.6).Transcripts, chat logs, disputes, statementsRead more
DORA
EU
Banks, insurers, investment firms, payment institutionsNon-production stores only anonymised, pseudonymised or randomised production data (RTS 2024/1774, Art. 16(5)).Statements, loan files, KYC packs, claimsRead more
CCPA and CPRA
California
Businesses meeting the CCPA thresholdsDeidentified information falls outside personal information only with reasonable safeguards, a public commitment not to re-identify, and contracts binding recipients (Civ. Code §1798.140(m)).Support tickets, emails, HR filesComing
GLBA Safeguards Rule
US
Financial institutions, including lenders and tax preparersA written security programme with a data inventory, access controls, encryption and secure development practices (16 CFR 314.4).Loan files, tax returns, statementsComing
NYDFS Part 500
New York
New York-regulated financial services firmsNonpublic information is in scope in every system: access privileges, encryption and asset inventory (23 NYCRR 500).Account files, claims, KYC packsComing
EU AI Act
EU
Providers of high-risk AI systemsTraining, validation and testing data must meet data-governance and quality criteria (Art. 10).Training and evaluation corporaComing
Subject access and FOI
UK and EU
Controllers answering DSARs; UK public authoritiesRelease the requester's own data and protect everyone else's (UK GDPR Art. 15; Freedom of Information Act 2000).Case files, emails, lettersRead more

Also often relevant: FCRA for consumer reports, state health-data laws, and rules from your own supervisor.

Free download

The test data controls checklist

One page per regulation: what an auditor will ask about your test environments, and the evidence that answers it. GDPR, HIPAA, PCI DSS and DORA.

Thanks. We'll send the checklist to your work email.
That didn't go through. Try again, or email team@synthesized.io.
What it covers
  • Every non-production environment, including user-built tools
  • Which documents, mailboxes and notes ride along with each refresh
  • The treatment for each category: mask, replace, redact or keep
  • Who approved the policy, and when it last changed
  • Exceptions: who signed them off and when they expire
  • The run record and re-scan for the last three refreshes

What they have in common

01
Minimise
Use real personal data outside production only when nothing else will do. Masked or synthetic data usually will.
02
Cover everything
The rules say personal data, not database columns. Documents, mailboxes, transcripts and notes are in scope too.
03
Prove it
Keep a record of every run: the policy, what it touched, what failed and how to reproduce it.

Questions about test data regulation

Is synthetic data outside these rules?

Usually, if it is generated rather than copied and doesn't reproduce real people. Check how it was made: synthetic data built badly from real records can still leak them.

What if more than one regulation applies?

Then all of them apply. A US bank's test copy can be in scope of GLBA, PCI DSS and NYDFS Part 500 at the same time, and an EU insurer's of GDPR and DORA.

Do these rules cover documents, or just databases?

Documents too. None of them limits itself to structured data: a statement, a scanned claim form or an email thread about a customer is personal data like the row it came from.

This page summarises the rules for planning purposes. It is not legal advice; check how they apply to you with your counsel or DPO.

The platform underneath, in production

Documents run on the same platform banks and insurers already trust with their test data.

Documents and text use the same policies, the same engine and the same deployment as these results.

Banking · test data80%less time to provision test data, with compliant data generated through the API inside CI/CD.Deutsche BankCase study
Insurance · security0security waivers across 14 priority applications: 28M+ rows secured, 100% referential integrity, test data 60% faster.Global specialty insurerCase study

Published results from database and synthetic data work on the Synthesized platform.

See your test data through a regulator's eyes

We'll take one environment, find the personal data in its tables and documents, and show what masking it would take.

Get a 48-hour proof