OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Security & Safety

Rizzo Pii

Rizzo Pii is a local document-anonymization application for Italian legal text. Its token-classification model identifies personal-data spans and replaces them with stable, type-aware placeholders, with a local dictionary that supports later restoration.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Rizzo Pii — OpenViST project profile in AI Security & Safety.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it before submitting Italian legal or other sensitive text to an external language model when you want to replace identifiers locally and retain a local mapping for restoration.

Who it is for

  • Legal professionals preparing documents for use with external language models.
  • Accountants, notaries and other users handling Italian documents containing personal data.
  • Developers seeking a local preprocessing step for a document-to-LLM workflow.

Documented uses

  • Detect personal identifiers in Italian legal text.
  • Replace sensitive values with consistent placeholders before sharing text with a language model.
  • Restore original values locally in a returned response using the saved mapping.
  • Run document anonymization through a desktop build or a Docker-based inference service.
Before you choose

Requirements & limits

The repository documents desktop builds for Windows, macOS on Apple Silicon and Linux, plus a Docker inference option. The source repository specifies an MIT license.

  • The model targets Italian text and documents 22 personal-data categories.
  • The workflow is reversible through a local placeholder mapping; the user must handle the separate external language-model query.
  • The Docker inference image is CPU-only and self-contained after building, but the documented build needs network access.
  • The repository describes the model as about 0.3 billion parameters and lists approximately 0.5 GB RAM for CPU use; its Docker instructions separately note about 1.2 GB for model loading in the container.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Build and start the local Docker inference service

Run the documented CPU-based inference service with its listening port bound to the local machine.

Before you start

  • Docker
  • Network access during the image build
  • A cloned copy of the Rizzo Pii repository

Steps

  1. Clone the repository and change into its directory.

  2. Build the image with the command below.

    Documented command
    docker build -t rizzo-pii .

  3. Start it with the port bound to localhost: the command below.

    Documented command
    docker run -d --name rizzo-pii -p 127.0.0.1:5005:5005 rizzo-pii

  4. Check the container health status; the service health endpoint reports 503 until the model is ready.

Expected outcome

A locally bound Docker inference service becomes available after the model has loaded.

Guide limitations

  • The documented image is CPU-only.
  • The build needs network access, while the running container is described as not requiring network access.
  • The documented command binds to localhost; publishing the port differently can expose the service to the LAN.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
Rizzo-AI-Academy/rizzo-pii · Official homepage title · retained in its source language
How we review sourcesSuggest a correction