Reconsider/rework detection of uploaded photos from lbpmod.me

[franbis]

Hello, I'm the creator of lbpmod.me, I just noticed the commit d375305 which blocks photos created specifically from that website. I find this damaging to the community as there are users who still create photos and, aside from this, Refresh already contains hundreds of mods created from lbpmod as well as photos as I removed the domain name from the photo subjects when the website upgraded to using CWLib, so that code is blocking nothing at all as of now.

That said, the only reason why you had to block photos based solely on the domain name (added by the website itself), is because you couldn't block them based on anything else as the rest is exactly like any other photo.

Also, in case you're wondering, the mods are created using CWLib and Workbench, both under the written authorization from ennuo from 2023-12-27.

[CrankyLandlord]

Pretty sure this was done for moderation purposes. Users were using this service to create and spam upload inappropriate photos with little to no effort

[jvyden]

Yes, this commit was made as a reaction to spam of unsafe content on our platform. The problem here is that it makes it very easy to mass-upload images to the public listing on our website. This has nothing to do with your website itself - rather the user behavior as a result of it.

Uploaded from a vanilla client or not, this is a common type of spam we see.

I fail to understand how blocking such photos is "damaging the LittleBigPlanet community" however. These photos are still allowed to be uploaded as part of level content, just not uploaded to the profile.

[franbis]

Yes, this commit was made as a reaction to spam of unsafe content on our platform. The problem here is that it makes it very easy to mass-upload images to the public listing on our website. This has nothing to do with your website itself - rather the user behavior as a result of it.

Uploaded from a vanilla client or not, this is a common type of spam we see.

I fail to understand how blocking such photos is "damaging the LittleBigPlanet community" however. These photos are still allowed to be uploaded as part of level content, just not uploaded to the profile.

Thanks to both for the quick reply.

My apologies, I thought you were talking about the website's behavior here.
This means the photo endpoint is used for the photo upload on the profile only, right?

Since the mass upload of photos is a problem, I will add a disclaimer in the "sticker editor" on the website.
I may add a way to track lbpmod.me files for servers later on to handle them if needed. As of now you could implement a block based on the item description, which contains the website domain name.

EDIT: If you're ok with this, I will edit the title of this issue to make it sound less problematic.

[Toastbrot236]

The item description is in the plan file, which would require deserialization in that case

[franbis]

I mean, if the issue is the spam, at that point it would make more sense to rate-limit the photo endpoint.
If, instead, as @CrankyLandlord said, the uploaded photos were "inappropriate", then it would make more sense to warn/ban the user.

[jvyden]

Well the problem isn't necessarily spam - it's the spam of the unsafe content itself. It's actually quite common for a single user to hit this endpoint normally as they upload photos in big batches.

I think for Refresh we're only interested in in-game photos taken with the actual capture tool so we can probably add heuristics for that.

[franbis]

Well the problem isn't necessarily spam - it's the spam of the unsafe content itself. It's actually quite common for a single user to hit this endpoint normally as they upload photos in big batches.

I think for Refresh we're only interested in in-game photos taken with the actual capture tool so we can probably add heuristics for that.

I understand it's not an issue related to the photos created from a specific website. What kind of unsafe content are we talking about in the context of photos? Scripts or inappropriate content?

[jvyden]

NSFW or sometimes hateful content.

[franbis]

NSFW or sometimes hateful content.

I see. Tho I'm thinking, if you know some users uploaded those photos, that means you know which users did that, which makes me wonder - why, instead of implementing heuristic, don't you just issue a warn/ban or add a report button?

[jvyden]

While we've been doing pretty well at deleting it, I'd rather reduce the opportunity it gets seen or posted in the first place. By doing heuristics to at least warn us of such images we can significantly reduce the chances it gets uploaded in the first place.

This is also why we already do content scanning on the upload endpoint and have alerts in our staff channels whenever something is flagged.

Whether we should do such heuristics I'm not going to argue - we are, because I have a duty to shield our users from such content.

edit: not to imply a report button on the website is a bad idea.