-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
We need a way of redacting offensive avatars and displaynames (SYN-566) #1458
Comments
Jira watchers: @ara4n |
To clarify: the redact API worked as intended (removing human readable contents of a join event: displayname and avatar). However, we have a seperate problem that the redacted data still leaks into the view of the room via the user's global profile data. I am proposing that redacting a this data on a join or profile event should also remove it from the per-room profile data (and that we should implement and uphold per-room profile data). This could be done as a synapse-specific feature for now, hence filing it here rather than blocking it on Matrix spec. -- @ara4n |
Related to element-hq/element-meta#1858 |
Okay, here's a decent solution for this that doesn't involve nuking your entire remote media repository: Example, let's start with the URL of the abusive image in question: First use the media_id (
Then use that filesystem_id to locate the remote file and all it's thumbnails:
Then zero each file and make it immutable:
It's now possible to do all of this easily with my moderation script, first run the 'Redact a room event.' section:
Then finally run the 'Delete and block a specific media.' section:
It's that easy! xD |
Submitted by @matthew:matrix.org
Currently the best you can do is redact the events which introduce the contents onto the timeline. However, the profile data remains intact and so the membership list etc continues as is. Given we have per-room profile data we should use it to redact the information from the room.
(Imported from https://matrix.org/jira/browse/SYN-566)
The text was updated successfully, but these errors were encountered: