By · Last updated 2026-05-29

Back to BlogGDPR & Compliance

Japan PPC APPI: AI Training Data Compliance

Japan's PPC enforces APPI 2022 amendments covering 2.4M Japanese enterprises. My Number 12-digit ID requires Verhoeff validation.

May 29, 202610 minute read
Japan PPCAPPI complianceMy Number detectionJapanese privacy lawAsia Pacific

Japan PPC and APPI: AI Training Data Compliance

Japan's PPC enforces APPI. The 2022 amendments changed the law more than any prior update. They added rules for pseudonymized records, cross-border transfers, and AI training sets. The PPC issued 45 rulings in 2024. It also published the first Japan-specific AI privacy guide that year.

If your firm trains models on Japanese text or holds Japanese user records, these rules apply now.

What the 2022 Amendments Changed

2.4 million Japanese firms had to update privacy rules and rework handling steps.

Pseudonymized information (仮名加工情報): A new middle class. It covers personal records with direct identifiers removed. Re-ID is still possible if you have a key. These records can move inside an organization without full consent. They cannot go to third parties. GDPR has no such class.

Anonymized information (匿名加工情報): Re-ID must be technically impossible. A qualified third party must confirm this. Japan's bar is higher than GDPR on this point. GDPR makes that review optional. APPI makes it required.

Cross-border transfers: Transfers to other nations must meet Japan's protection bar. The PPC keeps a list of approved countries. The EU is on that list.

AI training sets: PPC guidance from 2024 covered this directly.

  • Training sets must be fully anonymized or rest on a valid legal basis — usually consent.
  • The processing exception only applies if the model cannot identify people from its outputs.
  • LLM developers training on Japanese records scraped from websites must show a valid collection basis.

For a full view of cross-border alignment duties, see /legal/compliance.

My Number: Japan's National ID

My Number (マイナンバー) is a 12-digit national ID. Japan issues it to all residents. Foreign nationals get one too. The system has run since 2016. It covers tax, social security, and disaster response.

How the check digit works: My Number uses the Verhoeff method. It is a math-based error-check scheme. It is harder to build than Luhn — the method used for Swedish personnummer and Canadian SIN. Most European IDs use simpler modular math.

Why detection is hard: A scan for 12-digit strings will miss the mark. Dates, postal codes, and invoice codes all look the same. You need full Verhoeff logic to sort them out. Simple regex is not enough.

The PPC's 2024 review had a stark finding. 63% of generic NLP tools fail to detect My Number in Japanese records.

See how anonym.legal handles My Number at /entities.

Three Writing Systems at Once

Japanese uses Hiragana, Katakana, and Kanji all at once. Roman script appears in some contexts too. The same name may look different across records. Tools built for Latin-script text fail on Japanese without extra support.

What this means for name detection:

  • Japanese NER needs models trained on Japanese text. Use spaCy ja_core_news.
  • Japanese has no spaces between words. Word splitting is its own step. It needs Japan-aware tools.
  • Person names appear in Kanji with reading guides in Hiragana or Katakana. Tools must catch both forms.
  • Company names (会社名, 株式会社) need Japan-specific rules.

For NER across APAC languages, see /docs/faq.

Other Japanese ID Formats

Driver's license: 12 digits with a prefix code for the issue region. Codes are fixed — Tokyo is 10, Osaka is 62. The region part is checkable.

Passport: Standard ICAO format with Japan-specific issue rules.

Health Insurance Card (健康保険証): Symbol (記号) plus number. Format varies by insurer.

Residence Card (在留カード): For foreign residents. Format: two letters, eight digits, two letters. The Ministry of Justice issues these.

Japan–EU Data Transfer Status

Japan and the EU have had mutual adequacy since 2019. Personal records flow between the EU and Japan without extra steps. Japan is one of very few non-European nations with full EU adequacy.

The deal covers standard personal records. Sensitive health and criminal-history records need extra safeguards even under adequacy. Firms moving these records must log the added steps they use.

Review your transfer duties at /security-compliance.

Your Japan Compliance Checklist

Start here if you handle Japanese personal records:

  • My Number detection with Verhoeff check-digit logic.
  • Japanese NER with models trained on Japanese-script text — not Latin-script models.
  • Support for Kanji, Hiragana, and Katakana name forms plus reading guide variants.
  • Driver's license detection with region code checks.
  • Residence Card detection with MOJ format logic.
  • Health Insurance Card detection across issuer variants.
  • A valid legal basis for each AI training set that holds personal records.
  • Third-party review for any records classed as anonymized under APPI.
  • Extra safeguards for sensitive records moving under the EU–Japan adequacy deal.

See /docs/glossary for APPI term definitions used in this guide.

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.