presidio

This project targets the challenge of identifying and anonymizing personally identifiable information (PII) across text, images, and structured data using a combination of named entity recognition, pattern matching, and custom rule-based logic. Unlike simpler regex-only tools, it provides a modular framework that supports extensibility through custom recognizers and external models, allowing teams to tailor PII detection to specific business contexts. The inclusion of image redaction, including DICOM medical images, broadens its applicability beyond text-centric tools. With nearly 10,000 stars and an active commit as of mid-2026, it shows solid community interest and maintenance, though its growth has not accelerated recently. Machine learning engineers responsible for data privacy and compliance workflows will find it most useful. It is a poor fit if your primary need is lightweight or embedded PII masking in real-time data streams where lower-latency or more specialised libraries like Faker or scrubadub might be preferred.

Choose this framework when your project requires flexible, multi-modal PII detection and redaction that can be customised with domain-specific recognizers and integrated into Python or PySpark pipelines. If your needs are limited to text-only PII detection with less complexity, alternatives like spaCy with dedicated NER models or Amazon Comprehend may be simpler and more scalable.

It typically fits into data governance and compliance systems that process large volumes of sensitive text and images, such as automated document anonymization pipelines or healthcare data de-identification workflows involving DICOM images. Teams building privacy audit tools or preparing data for secure analytics often rely on it to automate redaction before data exposure.

Latest commit: 2026-06-28. Activity data is based on daily RepoPi snapshots of the GitHub repository.