Skip to main content

Alcatraz

Alcatraz ships 45 built-in recognizers covering 12 countries. 25 of them validate a checksum before reporting a match: Luhn for credit cards, ISO 7064 mod-97 for IBANs, Verhoeff for Aadhaar, mod-11 for the Brazilian documents (CPF, CNPJ, CNH, PIS) and others. A number with the right shape but the wrong check digit is dropped rather than masked. It also recognises secrets such as AWS access keys, JWTs and private keys. The full set, with ✓ marking the checksum- or format-validated types: Every one of these detects a language-independent structured identifier. An IBAN or a Thai national ID looks the same in any surrounding text, so the whole set is active regardless of the content’s language. The three statistical types below are English-language. On top of the built-ins, Alcatraz supports the same custom recognizers as Presidio: regular expressions, compiled as RE2 and therefore without lookaround or backreferences, and deny lists, matched on word boundaries. The authoritative list is in the alcatraz repository.

Entity types that need the NER model

These three are statistical, not pattern-based: they come from an ONNX model that runs in-process on the agent. The agent needs that model on disk, which means either an -alcatraz image tag or a mounted model directory. See Get Started. A rule that selects one of them is refused by an agent without the model; every other entity type above keeps working.

Microsoft Presidio

For updated fields support, visit Microsoft Presidio’s official documentation. Also, Microsoft Presidio supports custom fields to adapt to your needs. For more information, visit Microsoft Presidio’s official documentation.

Google Cloud Data Loss Prevention

Global

Credentials and secrets

The infoType detectors in this section detect credentials and other secret data.

Argentina

Australia

Belgium

Brazil

Canada

Chile

China

Colombia

Croatia

Denmark

France

Finland

Germany

Hong Kong

India

Indonesia

Ireland

Israel

Italy

Japan

Korea

Mexico

The Netherlands

New Zealand

Norway

Paraguay

Peru

Poland

Portugal

Singapore

South Africa

Spain

Sweden

Taiwan

Thailand

Turkey

United Kingdom

United States

Uruguay

Venezuela

Documents

In addition to its ability to scan and classify information contained within documents, Sensitive Data Protection can classify documents into multiple enterprise-specific categories. When combined with personally identifiable information (PII) inspection scan results, this classification can be useful for document risk assessment, policy enforcement, and similar use cases. Based on Cloud Data Loss Prevention.