Learn how Prompt Builder alerts you to potentially harmful content in a prompt template response.
Required Editions
| Available in: Lightning Experience |
| Available in: Enterprise, Performance, and Unlimited Editions with the Einstein for Platform, or Einstein or Agentforce for Sales or Service add-on, or Agentforce Foundations |
The Einstein Trust Layer scans the large language model (LLM) prompt responses for toxic language, including violent, abusive, hateful, or sexual language, imagery, or themes that can cause harm.
When toxic language is detected, the harmful content badge appears in the Response section. The badge displays a warning count and highlights the specific verbiage that triggered the alert.
If the Harmful Content badge appears, update and retest the prompt template to remove the toxic content. If the reassessed template no longer contains toxic content, the badge warning count returns to 0.