GenerativeAgent now offers customizable safety and scope guardrails, allowing you to define what’s considered “in-scope” and “safe” for your specific needs. While maintaining core safety protections, this tooling enables you to expand permissible topics and adjust default guardrails to better align with your business policies and requirements.

For detailed information about configuring and using safety tooling, visit our Scope and Safety Tuning documentation.

​
FAQs

  1. Will adding custom categories remove default protections?

    No, default safety and scope guardrails remain active. Your configurations help customize permissible interactions.

  2. How can I adjust the scope if GenerativeAgent is too conservative?

    Add or refine in-scope or input safety categories to expand acceptable topics, keeping standard safety intact.

  3. Why does the safety tooling require an explanation?

    Input safety requires an explanation to provide context for why certain inputs are deemed safe, helping GenerativeAgent accurately apply exceptions relevant to your specific needs.

Was this page helpful?

Pinned VersionsOverview