You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Seemingly irregularly, the chosen effect of moderation labels resets from 'Warn' to 'Off' or 'Hide' for me, for various labellers.
The official Bluesky Moderation Service seems to be affected as well, but only the click-through preferences on the labeller's page, not the adult content settings directly under Settings > Moderation.
To Reproduce
I am not sure how to reproduce this quickly. Due to the nature of the bug, I don't become aware of its effects unless I re-check the moderation settings.
Here's me relevant context:
I am subscribed to a lot of labellers, some of which have many labels.
I have most labels set to 'Warn'.
Expected behavior
The labels should remain set to 'Warn':
Screenshots
Instead, the settings seem to often reset to 'Off' or 'Hide' across various labellers:
(Reproduction, as I had just gone through the settings to fix them.
These are, however, the same settings it had reset to for the AI Imagery Labeller, which iirc are the defaults for it.)
Describe the bug
Seemingly irregularly, the chosen effect of moderation labels resets from 'Warn' to 'Off' or 'Hide' for me, for various labellers.
The official Bluesky Moderation Service seems to be affected as well, but only the click-through preferences on the labeller's page, not the adult content settings directly under Settings > Moderation.
To Reproduce
I am not sure how to reproduce this quickly. Due to the nature of the bug, I don't become aware of its effects unless I re-check the moderation settings.
Here's me relevant context:
Expected behavior
The labels should remain set to 'Warn':
Screenshots
Instead, the settings seem to often reset to 'Off' or 'Hide' across various labellers:
(Reproduction, as I had just gone through the settings to fix them.
These are, however, the same settings it had reset to for the AI Imagery Labeller, which iirc are the defaults for it.)
Details
Additional context
#3310 is related. This would be far less of an issue if I could not-trust labellers with hiding content.
The text was updated successfully, but these errors were encountered: