Activate offensiveness protection for generative AI

  • Versão de lançamento: Australia
  • Atualizado 12 de mar. de 2026
  • 1 min. de leitura

    • Turn on offensiveness protection to log and add the option to block offensive content in AI-generated text and conversations.

    Antes de Iniciar

    Role required: sn_generative_ai.nsa_admin

    Por Que e Quando Desempenhar Esta Tarefa

    Generative AI is probabilistic, which means that outputs are based on probabilities, and using the same input twice does not guarantee the same output. Some of the material generated by AI could potentially be undesirable because of toxicity, sexism, or other offensive sentiment. Now Assist Guardian enables you to log any material that is detected to be offensive. If you choose, you can also block offensive material so that users don't see the generated content. Instead, they see a message stating that offensive material has been detected and blocked.

    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.

    Procedimento

    1. Navigate to All > Now Assist Admin > Settings.
    2. In the side panel, select the Now Assist Guardian > Offensiveness tab.
    3. Go to the Available for you tab to see which workflows you can choose from.

      If you have any offensiveness guardrails already activated, they appear in the Active tab.

    4. Select Activate for the workflow that you want to enable offensiveness protection on.
    5. Select your impact detection.

      Now Assist Guardian logs when offensive content is detected or generated when offensiveness protection is activated. You can also choose whether you want to block the content from the user. If you choose to block the content, the user sees a standardized message explaining that offensive material has been blocked instead of what was generated.

      Offensiveness guardrail for Now Assist Guardian with option "log only" selected

    6. Select Save.

    Resultado

    Now Assist Guardian's offensiveness guardrail is enabled on your instance for the workflow you have selected.

    O que Fazer Depois

    You can enable offensiveness protection for all Now Assist applications that you have enabled on your instance. If you want to change your detection impact, you can select more options (More options icon.) in the list of active workflows and choose Edit.

    You can deactivate offensiveness protection for your workflow at any time by selecting more options and choosing Deactivate.