PII Detection in 23 European Languages
PII detection across 23 European languages
PrivacyPromptAI supports 23 European languages, detecting and cleaning PII in each with high accuracy. Our pattern recognition adapts to language-specific naming conventions, address formats, and phone number patterns.
Supported Languages
Aktive Sprache: DeutschKlicken Sie auf eine Sprache zum Wechseln
Click any language below to change it.
🇬🇧
English
English
🇫🇷
French
Français
🇩🇪
German
Deutsch
🇪🇸
Spanish
Español
🇮🇹
Italian
Italiano
🇵🇱
Polish
Polski
🇵🇹
Portuguese
Português
Coming soon
🇳🇱
Dutch
Nederlands
Coming soon
🇸🇪
Swedish
Svenska
Coming soon
🇩🇰
Danish
Dansk
Coming soon
🇫🇮
Finnish
Suomi
Coming soon
🇳🇴
Norwegian
Norsk
Coming soon
🇨🇿
Czech
Čeština
Coming soon
🇬🇷
Greek
Ελληνικά
Coming soon
🇭🇺
Hungarian
Magyar
Coming soon
🇷🇴
Romanian
Română
Coming soon
🇧🇬
Bulgarian
Български
Coming soon
🇭🇷
Croatian
Hrvatski
Coming soon
🇸🇰
Slovak
Slovenčina
Coming soon
🇸🇮
Slovenian
Slovenščina
Coming soon
🇪🇪
Estonian
Eesti
Coming soon
🇱🇻
Latvian
Latviešu
Coming soon
🇱🇹
Lithuanian
Lietuvių
Coming soon
How Detection Actually Works Across Languages
PrivacyPromptAI uses two detection layers:
- Universal pattern matching: emails, IBANs (with checksum validation), credit card numbers, IP addresses, US Social Security Numbers, UK/US postcodes, and dates. These formats follow consistent international structures, so detection accuracy doesn't depend on document language.
- Name detection: capitalisation and honorific-title matching (e.g. "Mr. Smith", "Dr. Jane Doe"). This works on any Latin-alphabet text but is not a trained, per-language model — it doesn't specifically parse Hungarian surname-first ordering or Spanish/Portuguese double surnames as distinct cases.
- Unicode support: text containing diacritics (ñ, ü, ø, ł, etc.) is processed normally, since detection runs on standard Unicode-aware JavaScript regular expressions.
PrivacyPromptAI does not currently include dedicated detection for country-specific national ID formats (e.g. Spanish NIE, Dutch BSN, Romanian CNP, Polish PESEL). For documents containing these, review the cleaned output manually, or use the Custom Keywords feature (Pro) to flag known ID formats specific to your documents.
Automatic Language Detection
PrivacyPromptAI automatically detects the language of your document. No manual selection needed!
You can also process multilingual documents, the tool will detect PII regardless of language mixing.
Don't See Your Language?
We're continuously expanding language support. Request your language!
Request LanguageFrequently Asked Questions. Language Support
Universal patterns (emails, IBANs, phone numbers, credit cards, dates) are detected the same way regardless of document language, since these formats follow consistent international structures. Name detection uses capitalisation and title-prefix matching, which works across Latin-alphabet text but isn't a separate pattern set per language — so a mixed-language document is handled consistently rather than switching between per-country logic.
The language selector currently affects the interface language, not detection logic — PrivacyPromptAI does not yet include dedicated pattern detection for country-specific ID formats such as Romanian CNP, UK National Insurance Numbers, or Polish PESEL codes. Universal patterns like email addresses, IBANs and credit cards are detected reliably regardless of language. If your documents regularly contain a specific national ID format, the Pro Custom Keywords feature lets you flag it manually.
Yes. Language support is expanded based on user demand. If you need a language not currently in our list of 23, please contact us → with the language name and any specific PII patterns your use case requires (ID formats, phone number conventions, postcode structure). We prioritise requests with clear use cases.
Text detection itself works on Greek and Bulgarian content, since processing runs on standard Unicode-aware regular expressions that handle Cyrillic and Greek character sets without extra configuration. Universal patterns (emails, IBANs, credit cards) are detected the same as in any language. Note that dedicated national ID pattern detection for these countries isn't currently implemented — see the full features list → for exactly what's detected.
Name detection is based on capitalisation and honorific-title patterns (e.g. "Mr. Smith", "Dr. Jane Doe"), not a trained cultural-naming model. It doesn't currently distinguish Hungarian surname-first ordering (e.g. "Kovács János") or Spanish/Portuguese double surnames (e.g. "García López") as separate cases — both would be flagged using the same general capitalisation rules. For cross-cultural documents, always review the cleaned output to catch anything the pattern-based approach misses.
Pattern-based detection (emails, IBANs, phone numbers, IP addresses) is equally accurate regardless of language, since it doesn't rely on linguistic training data. Name detection uses capitalisation and title-prefix matching rather than a per-language trained model, so it doesn't have the "stronger for major languages" tiering that a true NLP model would — but it also means results are consistent rather than uneven across smaller languages. Universal identifiers are caught reliably in every supported language. Contact us → if you're seeing consistent issues with a specific document type.
Yes. Changing the language in the dropdown updates the detection settings for the next cleanup run; it does not erase or modify any text already in the tool. Your cleaned output and original input text remain in place until you clear them. This means you can process a document in one language, switch to another, and run a second cleanup pass, useful for multilingual documents where you want to optimise detection for two language sets separately. Your language preference is also saved to localStorage so it persists across sessions.
Related Resources