Presidio — обезличивание персональных данных

★ 10.3K

Presidio is an open-source tool that finds personal information in data and de-identifies it. Reach for it when text cannot be sent to a cloud model or handed to a contractor as-is: before it leaves, the tool strips names, phone numbers, addresses, email, card and document numbers, IP addresses and other identifiers. De-identification can be done several ways: remove, replace with a label, encrypt so the original can be restored, or mask partially — keeping the last four digits of a card, for example. It works beyond plain text, detecting and covering data in images and tabular files. The recogniser set is configurable: add your own by regular expression or train one for national document formats. Install it as a Python library or run it as a container service, so it slots into a pipeline right before the model call.