I noticed a bit of panic around here lately and as I have had to continuously fight against pedos for the past year, I have developed tools to help me detect and prevent this content.
As luck would have it, we recently published one of our anti-csam checker tool as a python library that anyone can use. So I thought I could use this to help lemmy admins feel a bit more safe.
The tool can either go through all your images via your object storage and delete all CSAM, or it canrun continuously and scan and delete all new images as well. Suggested option is to run it using --all once, and then run it as a daemon and leave it running.
Better options would be to be able to retrieve exact images uploaded via lemmy/pict-rs api but we’re not there quite yet.
Let me know if you have any issue or improvements.
EDIT: Just to clarify, you should run this on your desktop PC with a GPU, not on your lemmy server!
As a test, I ran this on a very early backup of lemm.ee images from when we had very little federation and very little uploads, and unfortunately it is finding a whole bunch of false positives. Just some examples it flagged as CSAM:
- Calvin and Hobbes comic
- The default Lemmy logo
- Some random user’s avatar, which is just a digital drawing of a person’s face
- a Pikachu image
Do you think the parameters of the script should be tuned? I’m happy to test it further on my backup, as I am reasonably certain that it doesn’t contain any actual CSAM
This is normal . You should be worried if it wasn’t catching any false positives as it would mean a lot of false negatives would slip though. I am planning to add args to make it more or less severe, but I it will never be perfect. So long as it’s not catching most images, and of the false positives most are porn or contain children, I consider with worthwhile.
I’ll let you know when the functionality for he severity is updated
How do you even safely test scripts/tools like this 😵💫
I’d bet there’s a CSAM test image dataset with innocuous images that get picked up by the script. Not sure how the system works, but if it’s through hashes then it would be pretty simple to add that to the script.
based db0 releasing great tools and maintaining a great community
Opened a relevant issue: https://github.com/LemmyNet/lemmy/issues/3920
Thank you for this! Awesome work!
By the way, this looks easy to put in a container. Have you considered doing that?
I don’t speak docker, but anyone can send a PR
I’ll try it out today. I’m about to start my workday, so it will have to be in a few hours. Fingers crossed I can have a PR in about 16 hours from now.
deleted by creator
Hi db0, if I could make an additional suggestion.
Add detection of additional content appended or attached to media files. Pict-rs does not reprocess all media types on upload and it’s not hard to attach an entire .zip file or other media within an image (https://wiki.linuxquestions.org/wiki/Embed_a_zip_file_into_an_image)
Currently I delete on PIL exceptions. I assume if someone uploaded a .zip to your image storage, you’d want it deleted
The fun part is that it’s still a valid JPEG file if you put more data in it. The file should be fully re-encoded to be sure.
In that case, PIL should be able to read it, so no worries
But I could take ‘flower.jpg’, which is an actual flower, and embed a second image, ‘csam.png’ inside it. Your scanner would scan ‘flower.jpg’, find it to be acceptable, then in turn register ‘csam.png’. Not saying that this isn’t a great start, but this is the reason that a lot of websites that allow uploads re-encode images.
my pict-rs already re-encodes everything. This is already a possibility for lemmy admins
Good to hear they have that covered already. Looks like a great tool!
As @[email protected] stated. They’re still valid image files, they just have extra data.
Just going to argue on behalf of the other users who know apparently way more than you and I do about this stuff:
WhY nOt juSt UsE thE FBi daTaBaSe of CSam?!
(because one doesn’t exist)
(because if one existed it would either be hosting CSAM itself or showing just the hashes of files - hashes which won’t match if even one bit is changed due to transmission data loss / corruption, automated resizing from image hosting sites, etc)
(because this shit is hard to detect)
Some sites have tried automated detection of CSAM images. Youtube, in an effort to try to protect children, continues to falsely flag 30 year old women as children.
OP, I’m not saying you should give up, and maybe what you’re working on could be the beginning of something that truly helps in the field of CSAM detection. I’ve got only one question for you (which hopefully won’t be discouraging to you or others): what’s your false-positive (or false-negative) detection rate? Or, maybe a question you may not want to answer: how are you training this?
I’m not training it. Im using publicly available clip models.
The false positive rate is acceptable. But my method is open source so feel free to validate on your end
Acceptable isn’t a percentage, but I see in your opinion that it’s acceptable. Thanks for making your content open source. I do wish your project the best of luck. I don’t think I have what it takes to validate this myself but if I end up hosting an instance I’ll probably start using this tool more often myself. It’s better than nothing at at present I have zero instances but also zero mods lined up.
Any thoughts about using this as a middleware between nginx and Lemmy for all image uploads?
Edit: I guess that wouldn’t work for external images - unless it also ran for all outgoing requests from pict-rs… I think the easiest way to integrate this with pict-rs would be through some upstream changes that would allow pict-rs itself to call this code on every image.
You might be able however integrate with my AI Horde endpoint for NSFW checking between nginx and Lemmy.
https://aihorde.net/api/v2/interrogate/asyncThis might allow you to detect NSFW images before they are hosted
Just send a payload like this
curl -X 'POST' \ 'https://aihorde.net/api/v2/interrogate/async' \ -H 'accept: application/json' \ -H 'apikey: 0000000000' \ -H 'Client-Agent: unknown:0:unknown' \ -H 'Content-Type: application/json' \ -d '{ "forms": [ { "name": "nsfw" } ], "source_image": "https://lemmy.dbzer0.com/pictrs/image/46c177f0-a7f8-43a3-a67b-7d2e4d696ced.jpeg?format=webp&thumbnail=256" }'Then retrieve the results asynchronously like this
{ "state": "done", "forms": [ { "form": "nsfw", "state": "done", "result": { "nsfw": false } } ] }or you could just run the nsfw model locally if you don’t have so many uploads.
if you know a way to pre-process uploads before nginx sends them to lemmy, it might be useful
Exactly. If the pict-rs dev allowed us to run an executable on each image before accepting it, it would make things much easier
Now if you can make this work with mastodon, i’d be eternally grateful.😁
It’s software agnostic. So long as you’re storing you’re images in object storage, it should work
deleted by creator
