Trainable classifiers are used in data loss prevention and auto-labeling policies

Overview

Trainable classifiers are machine learning-based classifiers that recognize content by meaning and context rather than fixed patterns. Unlike sensitive information types that match predefined formats, trainable classifiers can identify unstructured content like strategic plans, financial reports, or HR documents. Using trainable classifiers in auto-labeling policies, and data loss prevention (DLP) rules, extends protection to sensitive business content that pattern-based rules can’t reliably capture.

Remediation

• Learn about trainable classifiers
• Get started with trainable classifiers

Run it

The check runs as part of the whole assessment, with the ZeroTrustAssessment module installed:

Install-Module ZeroTrustAssessment -Scope CurrentUser
Connect-Maester
Invoke-Maester -Pack ZTA -SkipBuiltIn

Leave out -SkipBuiltIn to run Maester’s own checks in the same run.

About this check

Severity	Medium
Effort to fix	High
Impact on users	Medium
Pillar	Data, AI
Category	Advanced Classification
Product	Purview
Feature	Data classification
Area	Data protection
Licence	EXCHANGE_S_ENTERPRISE
Tenant	Workforce, External

Product, feature, area and ATT&CK are Maester’s: the assessment does not record them, so Maester keeps them in the pack’s taxonomy.json.

A Microsoft check, part of the Zero Trust Assessment pack. Text from ZeroTrustAssessment 3.1.0, © Microsoft, MIT licence.