Build low-latency Vision AI applications using our new open-source Vision AI SDK. ⭐️ on GitHub ->

Stream logo
Start Coding Free
Moderation Certification Course

Notifications and reviewer alert configuration

This lesson explains how to configure notifications and reviewer alerts in Stream to ensure urgent harms like self-harm, child exploitation, and violent threats are addressed immediately. It covers how webhooks push moderation events in real time, how to route alerts into tools like Slack or PagerDuty, and how to assign the right reviewers for specific categories. You’ll also learn best practices for escalation paths, selective notifications, and regular audits to keep your moderation system responsive and reliable.

Why Notifications Matter

Not all flagged content is equal. While some cases can sit in the queue for later review, others, like self-harm, child exploitation, or violent threats, demand immediate attention. Notifications and webhooks give admins the tools to ensure these cases never slip through the cracks.

Webhook Notifications

A webhook is a way for Stream to instantly send moderation events to another system in real time. Instead of waiting for a moderator to check the dashboard, Stream “pushes” the event to wherever you configure it.

What Triggers Webhooks

Webhooks can be fired for:

  • New flagged messages (text, image, video)
  • Automated actions taken (flag, block, shadowblock, bounce)
  • User-level events (a user flagged or banned)

How They Work

  1. Admin sets up a webhook URL in the Stream Dashboard.
  2. Whenever a moderation event occurs, Stream sends a payload (JSON) to that URL with details like:
    1. Action taken (flag, block, etc.)
    2. Harm category (harassment, self-harm, PII, etc.)
    3. Severity level (text) or confidence score (media)
    4. User metadata (who posted the content)
    5. Message or media snippet
  3. The external system (e.g., Slack, PagerDuty, custom API) receives the alert and triggers the workflow you’ve defined.

Common Use Cases

  • Slack Alerts → Post a message in a #critical-alerts channel when CSAM, self-harm, or violent threats are detected.
  • PagerDuty Incidents → Trigger a PagerDuty alert for high-severity cases outside of normal business hours.
  • Email Notifications → Send an email to trust & safety leads for urgent categories.
  • Custom Dashboards → Route webhooks into your own moderation dashboard or BI tool for tracking.

Reviewer Assignments

Notifications don’t just need to go “somewhere” they need to go to the right person or team.

Admins can:

  • Assign specific reviewers to categories (e.g., Child Safety team for CSAM, Policy team for hate speech).
  • Create escalation rules (if unreviewed for X minutes, escalate to senior moderator).
  • Filter notifications so that only High and Critical severity trigger external alerts.

Best Practices for Notifications

  • Be Selective: Don’t notify on everything, reserve webhooks for urgent harms.
  • Define Escalation Paths: Make sure critical categories always have a human responder.
  • Integrate with Existing Tools: Deliver alerts into Slack, Jira, PagerDuty, or your existing workflow system.
  • Test with Staging: Always test notifications in a dev/staging app before enabling in production.
  • Audit Regularly: Check logs to confirm alerts were received and acted on.

Notifications and webhooks extend moderation beyond the dashboard. By configuring them for critical harms and assigning reviewers appropriately, you ensure urgent violations are acted on in real time, no matter where your moderators are.

Next, we’ll cover audit logs and reporting, how every action is tracked, how to demonstrate compliance, and how to use data to refine your moderation strategy.