Rough port from DO code to integrate AI image scanning into Refresh. This should help reduce the photo endpoint abuse we've been seeing from DT, while using the same existing infrastructure that works incredibly well.
This does not scan every uploaded texture as this would increase server load significantly. It will only scan under certain circumstances:
for uploaded photos
for images uploaded via the API
This also brings in the option to automatically restrict users who upload images with banned tags if configured. This will be left off on production while I gauge if this is going to false-flag for normal non-abuse activity.
Oh, this also introduces support for a staff-only webhook. This could be used for a non-persisted audit log if we wanted to use that as a solution for #565.
Rough port from DO code to integrate AI image scanning into Refresh. This should help reduce the photo endpoint abuse we've been seeing from DT, while using the same existing infrastructure that works incredibly well.
This does not scan every uploaded texture as this would increase server load significantly. It will only scan under certain circumstances:
This also brings in the option to automatically restrict users who upload images with banned tags if configured. This will be left off on production while I gauge if this is going to false-flag for normal non-abuse activity.
Oh, this also introduces support for a staff-only webhook. This could be used for a non-persisted audit log if we wanted to use that as a solution for #565.