By · Last updated 2026-03-20

Back to BlogGDPR & Compliance

English-Only PII Tools: GDPR Gap

A German Steuer-ID (11 digits with checksum) is structurally unlike a US SSN. French NIR numbers have 15 digits. Polish PESEL and Swedish Personnummer.

March 20, 20268 minute read
GDPR multilingual complianceSteuer-ID detectionFrench NIRSwedish PersonnummerEU PII identifier formats

English-Only PII Tools: The GDPR Gap

GDPR Has No Language Preference

GDPR covers personal data in any language. German, French, Polish, Swedish — all are covered equally. A missed Steuer-ID creates the same legal risk as a missed Social Security Number. The law does not care about language.

Most PII detection tools do.

The leading commercial and open-source tools were built for English text. Their entity detectors reflect this. They cover US Social Security Numbers, US driving licenses, and NANP phone formats well. Detectors for non-English national IDs are less accurate. They are less well kept up. They miss real identifiers more often.

For firms across EU member states, this creates a coverage gap. The tool says detection is complete. But non-English identifiers remain in the data. These are often the identifiers with the greatest GDPR exposure in certain countries.

Data authorities see this. Auditors look for it. A tool can work well on English records. But if it fails on German or French records, it is not compliant. A clean report does not change that.

National IDs Differ in Structure

The gap between English-centric tools and multilingual tools is not about adding more regex patterns. EU national identifiers are very different from each other. They need country-specific logic to be detected correctly.

German Steuer-Identifikationsnummer (Steuer-ID): 11 digits. It uses a checksum based on a Luhn formula variant. A generic SSN regex will not match it. A regex for any 11-digit number creates too many false positives in German documents.

French NIR (Numéro d'inscription au répertoire): 15 digits. The format encodes sex, birth year, birth month, and birth department. It also includes birth order and a 2-digit control key. The control key must be validated for correct detection.

Swedish Personnummer: 10 digits with a Luhn check digit. People born before 1990 use a + separator instead of -. That changes the format that must be detected.

Polish PESEL: 11 digits. It encodes birth date, gender, and a check digit based on weighted sums. Correct detection needs both format matching and checksum validation.

These are not variants of a common pattern. Each has a different length. Each uses a different check method. Each encodes data in a different position scheme. An English-trained NER model seeing a French NIR will not recognize it as a national identifier. It will ignore it or misclassify it.

The Practical Compliance Risk

Consider a compliance officer at a European BPO. They process data from Germany, France, Poland, and the Netherlands at once. Their tool reports successful PII anonymization.

But the result is not complete. Steuer-IDs in German records remain. NIR numbers in French records remain. PESEL numbers in Polish records remain. The tool's detectors for these formats are absent or too inaccurate.

Later, the dataset goes to analytics or to a research partner. The data still contains re-identifiable national identifiers. The GDPR issue does not appear in the tool's output logs. It surfaces when a data subject access request arrives. It may surface during a data authority audit. It may surface after a data breach.

Research comparing hybrid multilingual approaches against English-centric tools found clear results. Hybrid methods achieve F1 scores of 0.60 to 0.83 across European locales. English-only tools score near zero for non-English national ID formats.

See our GDPR compliance overview for how these gaps map to GDPR obligations.

What Full Coverage Requires

True multilingual PII detection for EU GDPR compliance needs three layers.

Language-native spaCy models provide semantic understanding in the language of the text. A model trained on German text knows that "Müller" is a common German surname. Models exist for 25 high-resource EU languages.

Stanza NLP models extend coverage to languages not in spaCy. This adds reach for more EU language communities.

Cross-lingual transformer models (XLM-RoBERTa) handle cross-language cases. A name in a French sentence is recognized as a person name. This works even if the engine was not trained on that specific name.

Regex with country-specific validation covers structured national identifiers. Steuer-ID, NIR, PESEL, and Personnummer each need their own checksum logic. This cuts false positives. Digit sequences that fail country validation rules are filtered out.

The gap is structural. Adding word lists or more regex patterns gives only minor improvement. Building in EU identifier coverage from the start is the only reliable approach.

Check Your Current Tool

Ask your vendor for F1 scores on German, French, Polish, and Dutch records. "Supports multiple languages" often means the tool uses translation first. That is not native scanning. GDPR compliance requires native scanning.

Test with real national ID samples. Build a short test set with 10 examples of each ID type in your operations. Steuer-ID, NIR, PESEL, Personnummer. Check detection rates. This is faster than a full F1 test and shows gaps quickly.

See our security and compliance page for how anonym.legal addresses these requirements. For entity type definitions, visit the entities reference.

Sources

Ready to protect your data?

Start anonymizing PII with 285+ entity types across 48 languages.

About this page

We update this page when our platform or the law changes.

Read our founder note for how we work.

Each change shows up in the timestamp at the top.

Related reading

We follow these rules

  • GDPR (EU 2016/679).
  • ISO/IEC 27001:2022.
  • NIS2 (EU 2022/2555).
  • HIPAA safe harbor under 45 CFR § 164.514(b)(2).

Our promise

We do not sell your data.

We do not train models on your text.

We store your files in Germany.

You can delete your account at any time.

You own your work.

Where we run

Our servers live in Falkenstein, Germany.

We use Hetzner. They hold ISO 27001 certification.

All data stays in the EU.

Backups run every day.

Need help?

Email support@anonym.legal.

We reply within one business day.

How we test

We run a full check suite on every release.

Each surface gets its own sweep script and report.

Human reviewers spot-check the output each week.

We track recall and precision on a labelled set.

Bad runs block the deploy.

What we never do

  • We never sell your information to third parties.
  • We never train models on what you upload.
  • We never keep your work after you delete it.
  • We never share keys with any outside firm.
  • We never run ads inside the product.

Plans in plain words

We sell credits, not seats.

One credit covers one short job.

Long jobs use a few credits each.

You can top up at any time.

Unused credits roll over each month.

Read the plans page for current rates.

Who built this

A small team of engineers and lawyers built this.

We ship from Europe and work in the open.

Our founder note spells out why we started.

Where to start

How the parts fit

A browser add-on cleans text inside Chrome.

A Word plug-in handles drafts in Office.

A small desktop tool works on whole folders.

An agent protocol link feeds large models safely.

All four share one core engine and one rule set.

Words from our team

We started this work after a lunch about cookies.

One friend kept getting odd ads on her phone.

We asked why a court file leaked through a draft.

We sketched the first build on a napkin that week.

By month three we had a tiny demo for a friend.

She used it on her first case the next day.

Common questions we hear

Can the tool read scanned PDFs? Yes, with OCR.

Does it work on long files? Yes, in small chunks.

Can I roll my own rule set? Yes, save it as a preset.

Does it run offline? The desktop build runs offline.

Do you keep my files? No, the cloud build wipes after each run.

Will it learn from my work? No, we never train on inputs.

A short tour of the workflow

Upload a file or paste a snippet of prose.

Pick the entities you want gone from the draft.

Choose a method: replace, mask, hash, encrypt, or redact.

Press run and watch the side panel show each hit.

Skim the result and tweak any rule that misfired.

Save the cleaned file or send it to a teammate.