Policies & Rules

Policies control what the system watches for and how it responds. They connect AI detections to alerts.

Key Concepts

• Policy — A named collection of rules. Can be a family default or child-specific.
• Rule — A single condition: “When category X is detected, do Y with severity Z.”
• Category — A type of harmful content (e.g. cyberbullying, grooming).
• Action — What happens when a rule matches: immediate alert, digest, or silent log.
• Severity — How serious: low, medium, high, or critical.

How It Works

Screenshot → AI Analysis → Detections → Policy Rules → Alerts

1. The AI analyses a screenshot and reports zero or more detections
2. Each detection has a category and confidence score
3. The system finds the applicable policy for the child
4. Each detection is matched against the policy’s rules
5. Matching rules create alerts with the configured action and severity

Without a policy (or with an empty policy), detections are recorded in the activity log but no alerts are generated.