Follow along with our weekly product updates for Stream Chat, Feeds, Video, and Moderation.
Chat
Updates — March 31st, 2025
-
Drafts—Coming Soon
Draft message support is rolling out this week for both Android and SwiftUI.
-
iOS SDK Fixes
- Fixed an issue where draft local attachments were erased when the draft updated event was triggered.
- Resolved a bug where background tasks weren’t consistently running in
IOSBackgroundTaskScheduler
. - Composer content now clears properly when a draft deleted event is triggered.
- Updated
ChatChannelListItemView
to useColorPalette.text
for thetitleLabel
.
Feeds
Updates — March 31st, 2025
-
Audit Logs (Enterprise Preview)
- We’re rolling out audit logs to select enterprise customers. While documentation is not live, here’s a sneak peek of the unpublished docs:



Video
Updates — March 31st, 2025
-
iOS SDK
- Fully compatible with Swift 6.
- Fixed an issue where Picture-in-Picture mode stayed active even when the app was in the foreground.
- Updated WebRTC dependency to resolve a crash that could occur on Hang Up in certain scenarios.
- The
VideoRendererView
for the local participant now flips only the front camera feed—no longer flipping the back camera. - Better handling for blocked users
-
Android SDK
- Prioritized software decoding for specific codecs to fix an issue where all key frames were dropped, resulting in video loss.
- Improved handling of call ended events.
Moderation
Updates — March 31st, 2025
-
LLM-Powered Moderation (Beta)
- We now support full LLM-backed moderation for dramatically improved accuracy and contextual understanding. This update allows your moderation to go beyond keyword filtering and understand meaning across conversations. Contact us to enable it on your app.
-
Ruby SDK: Moderation API Support
- Now available in v3.12.0.
-
CSV Export for Moderation Logs
- You can now export moderation logs as CSVs directly from the dashboard. We received many requests from moderators who maintain their analytics in spreadsheets, so we listened. You'll find a new export button on the dashboard that lets you analyze the logs however you want.

-
Improved Test Policy Visibility
- The dashboard now displays all detected classifications, not just ones with configured actions. That means clearer visibility into what your AI moderation is detecting—no more guessing.
- Previously, the dashboard only showed moderation classifications if you had configured specific actions for them—making it tricky to tell whether AI moderation was working or just taking a coffee break. That led to some understandable confusion.
- Now, we’ve made things much clearer: all classifications are shown, whether or not they’re tied to an action. It’s full transparency into what your AI is detecting—no more guessing games.
