Company
A swiss-based company
Founded in Switzerland.
Artificial Intelligence Suisse SA, PO 280, Delemont, Switzerland.
Founded in Switzerland.
Artificial Intelligence Suisse SA, PO 280, Delemont, Switzerland.
ai4privacy
Collection
New release
The world's largest open multilingual PII masking corpus. 3M+ synthetic examples across 30 languages spanning Europe, the Americas, and Asia-Pacific, purpose-built for training privacy-preserving NLP models in a truly global setting.
The Asia-Pacific expansion is delivered together with VNCyberS, pioneering data protection and cybersecurity in Vietnam, bringing regional language expertise and on-the-ground privacy compliance to the release.
pii-masking-openpii-1.5m
The expanded open-source core, now covering Asia-Pacific languages alongside Europe and the Americas across 30 languages.
pii-masking-work-pwi-400k
Work & HR Information (PWI): job titles, organisations, salaries, document numbers, and employment identifiers.
pii-masking-financial-pfi-400k
Financial Information (PFI): IBAN, account & card details, balances, crypto wallet addresses, and insurance policy numbers.
pii-masking-location-pli-400k
Location & Travel Information (PLI): geo-coordinates, addresses, airport & station codes, and vehicle and travel identifiers.
pii-masking-health-phi-400k
Health & Medical Information (PHI): diagnoses, medications, test results, allergies, hospital names, and medical record numbers.
pii-masking-digital-pdi-350k
Digital Information (PDI): usernames, passwords, API keys, MAC addresses, device IMEIs, OTPs, and user agents.
openpii-masking-micro-100k
100K open-core samples for evaluating PII detection across all 30 languages.
openpii-masking-nano-1k
1K open-core samples for rapid iteration, CI/CD pipelines, and quick provider comparisons.
pii-masking-micro-100k
100K full-taxonomy samples — the largest benchmark slice across all 30 languages.
pii-masking-mini-10k
10K full-taxonomy samples for quick model and commercial-API evaluation.
pii-masking-nano-1k
1K full-taxonomy samples for the fastest smoke tests across all 30 languages.
7 new languages
23 locales
North & South
{
"source_text": "本日の集合場所は 射水市 円池 の 昼場 6-28-20、郵便番号は 520-2111 です。",
"masked_text": "本日の集合場所は [CITY_1] の [STREET_1] [BUILDINGNUM_1]、郵便番号は [ZIPCODE_1] です。",
"privacy_mask": [ { "value": "射水市 円池", "label": "CITY" }, { "value": "昼場", "label": "STREET" }, { "value": "6-28-20", "label": "BUILDINGNUM" }, { "value": "520-2111", "label": "ZIPCODE" } ],
"language": "ja", "region": "JP", "script": "Jpan"
}Train NER models to detect and classify PII entities with pre-computed mBERT-compatible BIO labels.
Build production-grade anonymization pipelines compliant with GDPR, the EU AI Act, PDPA, APPI, and PIPA.
Fine-tune large language models for privacy-aware text generation and redaction across languages.
Get access to the full 3M dataset including all industry-specific components and Asia-Pacific coverage, with commercial licensing for your organization.