Policy Configuration
Every policy in the catalog is fully configurable, allowing you to tailor enforcement behavior to your organization's risk tolerance and operational requirements.
Enforcement actions — Each policy supports one of the following:
| Action | Behavior |
|---|---|
| Block | Prevents the content from being delivered. Returns a configurable block message to the user. |
| Redact (Mask) | Replaces sensitive information with masked placeholders. The message is delivered with sensitive data removed. Available for Privacy policies. |
| Warn (Detect) | Flags the content for review without interrupting the conversation. The content is delivered unchanged — nothing is blocked or masked — but the violation is logged and surfaced for analysis. Available for every policy type, Privacy included, so you can monitor a policy before enforcing against it. |
| Log | Records the event for analytics and audit purposes with no user-visible action. |
Rule-based actions
The Action Mode setting chooses between Basic, one action for the whole policy, and Rule Based, a different action for each set of conditions. Each rule combines conditions on the application, message type, media type or a custom field with AND or OR (each condition can be negated with If not), and the rules are checked in order. An Else action applies to content no rule matches, and can be Do nothing. On a Privacy policy, a rule or the Else can also mask. Rules apply from sets the confidence level a score must reach before any rule or the Else applies; below it, nothing is detected. A rule or Else that blocks carries its own block message. In Basic mode, the block message is labelled Condition response.
A policy that has rules or an Else shows a Rule-based pill on Enforcement Policies instead of a single action. Switching back to Basic clears the rules.
Message type targeting — Scope each policy to specific parts of the conversation:
- All — Applies to both user inputs and AI responses.
- Only Prompt — Applies only to user messages (input guardrails).
- Only Response — Applies only to AI-generated responses (output guardrails).
Keyword augmentation — Supplement model-based detection with keyword rules for additional precision.