I don’t know if you need this info, but I was pretty disturbed to see unexpected child pornography on a casual community. Thankfully it didn’t take place on SLRPNK.net directly, but if anyone has any advice besides leaving the community in question, let me know. And I wanted to sound an alarm to make sure we have measures in place to guard against this.
Unfortunately I saw it as well while scrolling, and reported it. What’s the motivation behind posting fucked up shit like that?
I don’t know the specifics, but trolling is trolling. It’s experimenting with ways of breaking things. Not only do they probably find it funny, but if this isn’t handled it can kill the platform. If they saw that Lemmy.World was defederated and shut down, that would make their day.
The point is that we need basic security measures to keep Lemmy functioning. I don’t think this is just an issue of moderator response times. We need posts like that to get deleted after 10 people downvote it, and we need limits on how easily new accounts can get into everyones’ front page feeds.
It should be reports and limited to users with some form of track record on the platform. So posted some time earlier, has gotten X likes, account age and similar measures to make sure it is not problematic.
Downvotes are a bad measure. They are often just done by somebody disagreeing with a post, which often are not exactly a problem. Also 10 is really low, when something really takes off. On the c/meme half the posts have more then 10downvotes, but nothing is really all that bad.
The best suggestion I have seen is to have a specific report category for CSAM. If a post is reported for CSAM x number of times, the post is hidden for moderator review. If it is a false report, the mod bans the reporting accounts.
Another issue is that post links can be edited. Trolls will definitely use this feature for abuse.
Ranking algorithms need to be adjusted so that if a post is removed like this, and then restored, it gets the same number of views it otherwise would have. Without that, a user-interaction driven automatic removal will get abused at scale.
You can see below in this thread herein that someone got >10 downvotes for a perfectly reasonable and civil post by people who merely disagreed with their comment. Automatic censorship would be overly interventionalist. I would not want to participate in a forum that auto-censored like that. Downvotes did their job – pushed the msg low on the page for reduced visibility.
Your idea with restricting who can get into front page, is a really great idea! I will write it down for a project of ours.
If the same trolls got 10 accounts, they could find some other way to exploit the security gap, and also delete any posts warning about it.
Maybe it would help if communities could turn off image uploading? I mean asklemmy doesn’t hardly ever has a reason for there to be a picture. Communities that need it of course would still need other security measures.
The problem with an automatic delete is that it’s just as exploitable. Anyone can set up 10 accounts on various hosts, or even on one host, and gain the power to instantly delete anything they like
The alternative is requiring 24-hour moderation, which isn’t really feasible unless moderators are paid employees, or just having to deal with posts staying up until a moderator/admin comes online and can sort them out. Communities can obviously try to have a mod team comprised of people from a range of time zones to increase coverage, but aiming for 24-hour coverage would make most mod teams far larger than is particularly necessary for the size of most communities at the moment.
Posts being removed and flagged to moderators for review if a certain report threshold is met is the best middle ground for a community-run, non-commercial forum. Sure, someone can set up 10 (or however many the threshold is set at) accounts and report a post on all of them to have it removed until a moderator is online, but is it really worth it to go through that effort just to get a post taken down for a couple of hours before it gets reinstated?
It’s the best way to allow the community to self-moderate, I think, rather than requiring all the moderation power be in the hands of those with a moderator role.
On top of what’s been said, it should probably not delete but just hide it, so that mods could still re-approve it in case of mistakes
Removed by mod
As a mod of a few communities I’d just turn off public posts and contact the admins to block any trouble accounts, luckily I haven’t seen anything yet
Can you explain? I don’t know what it means to turn off public posts?
As community moderator you can apply a setting that means only mods can post on that community and it’s not possible for anyone to post anything.
Removed by mod
Mods and admins care, but we’re not all online all the time.
You are right on the point! We are all do this in our free time and we are searching for admins that are free in a timezone we still dont have covered yet.
We are open if someone is interested in assisting us, just hit us with an email with some details about you and when you can be active on lemmy.world.
That’s pretty shocking.
What tools are available to us to manage this?
The best tool that is currently available is lemmy-safty AI image scanning that can be configured to check images on upload or regularly scan the storage and remove likely csam images.
It’s a bit tricky to set up as it requires an GPU in the server and works best with object storage, but I have a plan to complete the setup of it for SLRPNK sometimes this year.
This is probably the best option; in a world where people use ML tools to generate CSAM, you can’t depend on visual hashes of known-problematic images anymore.
