Skip to main content

Policies

Control Center for your AI Agents​

tip

You can update policies for each API Key in the ZenGuard Policies page.

ZenGuard Trust Layer provides detectors to help AI agents detect various threats, misuse and private data exposure. The configuration of the detectors to be run is controlled centrally by the ZenGuard Policy.

Currently each Policy is per an API Key.

The Policy is designed to give you a single place to set up configuration for the detectors per each AI agent or AI application. You can enable/disable detectors and setup applicabale controls per detector.

The default Policy enables all detectors with sensible default configuration that works for most use cases.

Detectors​

You can enable/disable following detectors:

  • Prompt Attacks
  • PII
  • Intended Use: Allowed and Banned Topics
  • Secrets
  • Keywords

You can setup granular controls for following detectors:

  • PII:

    • Enable/Disable PII detection per entity type.
    • Select PII Entities redaction type.

  • Intended Use:

    • Add/Remove allowed topics
    • Add/Remove banned topics

  • Keywords:

    • Add/Remove keywords

Contents