gws-modelarmor

Prompt sanitization filters user-generated content through policy-based rules before it reaches AI systems. You can define reusable templates to clean or constrain inputs and outputs, ensuring safety checks are enforced automatically across model prompts and responses.

Template-driven content moderation enforces safe content policies by defining reusable templates for content rules. This allows you to systematically block disallowed phrases and redact sensitive data across AI prompts and responses before sending them.

You can filter AI responses by running template-aware safety checks via CLI integration. This applies preconfigured armor templates to clean or constrain outputs, redacting sensitive data and blocking disallowed phrases before responses are sent to users.

Yes, policy-based filtering works with custom safety templates by allowing you to define and enforce specific content rules. These templates automate the enforcement of safe content policies, reducing risk and maintaining compliance for AI systems.

Yes, you can use prompt sanitization for support chatbots to block disallowed phrases and redact sensitive data. The template-driven safety policies filter user-generated content before it reaches the AI, ensuring responses meet safety standards.

Automated content moderation relies on predefined template rules for filtering, meaning its effectiveness is bounded by the specific policies configured. It requires properly maintained safety templates to consistently identify disallowed phrases and sanitize sensitive data.