Use ML-Based Document Classification for Efficient DLP Policy Management

Introduction

This document describes how to use ML-based document classification (Built with Meta Llama 3) to simplify and enhance data loss prevention (DLP) policy management.

Overview

ML document classification improves content matching and makes DLP policy management easier. You can select from predefined, pre-trained document types, eliminating the need to build complex data classifications from scratch based on terms and patterns.

How to Use ML-Based Document Classification

• ML-based document classification is available as a built-in identifier when you create a data classification.
• All supported ML-based identifiers appear in the  ML Built-in Identifiers section during data classification setup.

26221783199892

Supported Document Types

You can use ML-based classification for these document types:

• Consulting agreements
• CVs and resumes
• IRS forms
• Medical power of attorney
• Non-disclosure agreements (NDA)
• Partnership agreements
• Stock and US patents

Supported Languages and Regions

The initial release supports US English language documents only. Support for additional languages and regions is planned for future updates.

Multimode DLP Support

ML-based document classification works with both Realtime DLP and SaaS API DLP. It is compatible with all DLP supported file types.

Related Resources

Refer to Secure Access and Umbrella documentation for guidance on using ML-based data identifiers in data classifications and DLP rules:

• Secure Access: Built In Data Identifiers
• Umbrella: Built In Data Identifiers