Copy and Customize a Built-In AI Guardrails Data Classification

The system provides three different built-in AI guardrails data classifications you can use in AI guardrails rules. These are:

Security Guardrail

A collection of data identifiers that protect against prompt injection attacks, ensuring that prompts are safe for the AI model and will not cause the model to perform inappropriate activities such as deleting critical data, exposing sensitive data, or granting unauthorized access to resources in your environment.

Safety Guardrail

A collection of data identifiers that protect against transmission of inaccurate information or information that could be used in a harmful manner.

Privacy Guardrail

A collection of data identifiers that prevent the AI application from disseminating personal Information Identifiers (PII) that may identify an individual, such as financial account numbers, medical records, passport or government identification numbers, or credit card numbers.

You can create custom AI guardrails data classifications based on any of these built-in AI guardrails data classifications. Simply copy the desired built-in AI guardrails data classification, make changes to the copy, and save it as a custom AI guardrails data classification.

The building blocks for AI guardrails data classifications are AI guardrails data identifiers. The AI guardrails data identifiers selected for an AI guardrails data classification determine the type of data for which rules using that AI guardrails data classification will scan. Built-in AI guardrails data classifications have AI guardrails data identifiers already selected for them; you can produce your own customized versions of these classifications by removing selected identifiers or adding other identifiers.

Prerequisites

• Full Admin user role. For more information, see Manage Accounts.

Procedure

1. Navigate to Secure > Settings > Data Loss Classification > AI Guardrails Classifications.

2. Expand one of the built-in AI guardrails classifications and click Copy.

3. Give the new AI guardrails classification a meaningful name and description.

4. Under Included Data Identifiers deselect pre-selected AI guardrails data identifiers as needed.

5. Under Include Data Identifiers, you may change the Boolean operator to separate the identifiers included in the classification:

• OR—At least one of the data identifiers selected must match during rule evaluation.
• AND—All of the data identifiers selected must match during rule evaluation.

6. Under Include Data Identifiersexpand Built-in Data Identifiers and select additional built-in AI guardrails data identifiers to include in the data classification.

7. Click Save.

Your new classification is listed on the Data Classification page under AI Guardrails Classifications, and will be available to you when you add an AI guardrail rule.

---

Create an AI Guardrails Data Classification< Copy and Customize a Built-In AI Guardrails Data Classification > Delete or Edit an AI Guardrails Data Classification