gws-modelarmor-sanitize-response

Template-based sanitization applies a Model Armor template to outbound model responses to remove or mask unsafe content. It enforces safety and compliance before delivering outputs from AI assistants, chatbots, and integrated apps to end users.

Sanitize outbound content in an AI chatbot by applying a selected Model Armor template to the model output. The skill validates inputs, processes the template, and returns sanitized content to prevent leakage of sensitive information.

Apply a Model Armor template to model outputs whenever an integration exposes raw model responses to users. It is needed to enforce safety guidelines, mask unsafe content, and prevent sensitive information leakage in outbound content.

Yes, Model Armor templates work with existing AI assistant integrations that expose model outputs. The skill processes the template against outbound content, validates the inputs, and returns safe, compliant responses for chatbots and integrated apps.

The best way to prevent sensitive information leakage in model outputs is template-based sanitization. Applying a Model Armor template to outbound responses removes or masks unsafe content, ensuring the delivered response meets safety and compliance guidelines.