Use AI Guardian features in AI Admin Center

  • Release version: Zurich
  • Updated July 30, 2026
  • 1 minute to read

    • Use AI Guardian features in the AI Admin Center workspace to detect offensive content, prompt injection attacks, and sensitive topics in generative AI interactions.

    Before you begin

    The following applications must be installed before performing this task:

    Role required: sn_na_center.nac_admin

    About this task

    Follow these steps to use AI Guardian capabilities from AI Admin Center.

    AI Guardian provides safety and governance controls for AI-generated content. It monitors AI interactions for potentially harmful, offensive, or policy-violating content.

    In AI Admin Center, the integration of AI Guardian includes multi-tabbing support for working with safety and governance controls without leaving the application context.

    For more information on AI Guardian, see AI Guardian.

    Procedure

    1. Navigate to All > AI Admin Center or Workspaces > AI Admin Center.
    2. Select Admin (Admin icon. ) in the side navigation bar.

      The Admin tab opens showing AI Admin Hub options.

    3. Select one of the options under AI Guardian to configure.

      AI Guardian provides three guardrails. Each guardrail has a different scope.

      GuardrailDescription
      Prompt injection detection

      This guardrail attempts to override LLM instructions or expose restricted information. It applies to all generative AI applications and features.

      Select Prompt injection to open the Prompt injection tab.

      For more information on how to configure this guardrail, see Configure prompt injection attack protection.
      Offensiveness detection

      This guardrail detects offensive or harmful content in AI inputs and outputs. It applies to specific generative AI skills and workflows.

      Select Offensiveness to open the Offensiveness tab.

      For more information on how to configure this guardrail, see Activate offensiveness protection for generative AI.
      Sensitive topic filters

      This guardrail filters subjects not suited for AI responses, such as workplace safety or employee compensation. It applies to Virtual Agent conversational skills only (available for HR Service Delivery and Customer Service Management).

      Select Sensitive Filters to open the Filters tab.

      For more information on how to configure this guardrail, see Configure sensitive topic filters.