Moderation

Stream Chat supports moderation actions like flagging, muting, banning, and shadow banning. For backend moderation concepts and APIs, see the moderation docs.

You can also moderate flagged messages and users in the moderation dashboard.

Best Practices

• Align dashboard moderation actions with your in-app UX expectations.
• Surface clear feedback when a message is bounced or blocked.
• Treat message.moderation_details as the primary source of truth for moderation state when it is present, but be aware moderation state can also surface on message.moderation.
• Keep resend actions gated to prevent repeated policy violations.
• Customize MessageBounce only if you keep the same moderation semantics.

Moderation Categories

For each channel type, you can enable AI moderation, consisting of 4 moderation categories:

• Semantic Filters
• Commercial Spam
• Platform Circumvention
• Blocklists

For each category, configure the action applied when a message matches:

The available actions are:

• Flag
• Block
• Bounce
• Bounce then flag
• Bounce then block

React Native SDK Handling

In the React Native SDK, check message.moderation_details. It’s undefined for regular messages.

If moderation blocks a message, message.type is error and the text contains Message was blocked by moderation policies.

If moderation bounces a message, message.type is error and moderation_details is defined.

UI Components handling

The SDK handles bounced messages by default and shows a visual indicator.

Press/long-press a message to access default actions like edit, delete, and resend.

You can customize bounced message actions by passing a custom MessageBounce to Channel.