Configure prompt injection attack protection

  • Freigeben Version: Australia
  • Aktualisiert 12. März 2026
  • 1 Minute Lesedauer

    • Activate or deactivate prompt injection attack protection for AI-generated text and conversations.

    Vorbereitungen

    Role required: sn_generative_ai.nsa_admin

    Warum und wann dieser Vorgang ausgeführt wird

    Prompt injection attacks are a type of cybersecurity attack where someone tries to override the initial instructions of an LLM to cause unintended behaviors. Now Assist Guardian can detect and log these attack attempts, and you can choose whether you want to block the AI-generated response after the attack has been attempted.

    See Now Assist Guardian for more information.

    Logs can be exported for review. For instructions on how to do so, see Export Now Assist Guardian logs.

    Prozedur

    1. Navigate to All > Now Assist Admin > Settings.
    2. In the side panel, go to Now Assist Guardian > Prompt Injection.
    3. Select the toggle to change the active status of prompt injection attack protection.

      Prompt injection attack protection is enabled by default, which means Now Assist Guardian logs attempts unless you turn it off.

    4. Wahlweise: Under Detection impact, select the options icon (Options icon.) and then choose Edit to change the detection impact.

      You can choose whether prompt injection attacks are blocked as well as logged.

      Prompt injection protection detection impact selection card with "log only" selected

    Ergebnisse

    You have configured whether prompt injection attack protection is enabled on your instance and determined what you want Now Assist Guardian to do in case of an attempt.