(untitled)

there is a lot of CP on this board that gets posted by spamming bots. to my knowledge, lainchan is non-scrapable. are there alternative ways of user compromise in this case? it has to be including the picture link-click capture and occasional overviews of lainchan by feds, as well as anonymous tip possibility creation dont click

fvcking hell i can't unsee this

I want to blame this on the feds too, but even as a poster of 5 years I barely even understand who Kalyx and Appleman are, two very significant figures in this site's administration. I haven't been able to find much info even when I was active on IRC years ago. I can understand the need for anonymity considering this site's security culture though. There's so already a thread on this if you don't know, >>>/q/12273

>>6918 I just report it. Obviously logging and parallel construction is always a possibility. You start doing something they don't like and they try to frame you.

>>6918 We know, there's already like 3 threads on this subject. I have reported the website they used to one service a few times now, but nothing has been done, maybe they're out of jurisdiction I don't know. The other report services all have Cloudflare, javascript, and live captchas, and I don't want to deanonymize myself so I can't report them to somewhere like the FBI for instance. Anyway this is why you should torify everything with something like Whonix :D, then you don't need to worry as much when scumbags hurl cp at you. Here's the two other threads that went unmentioned. >>75372 >>77332

>>6923 would you be able to say how exactly would you torify lainchan with Whonix? given that most tor nodes are bought off by the US government in sybil

>>6918 Hi, former pedo here. I feel regret and shame for what I have been before when I was younger. In this I have thought of a simple technological advancement that I think would help all websites today. Issue with moderation today is that it's manual process that usually takes a long time and can cause mental torture to those moderating these websites. I have been toying around with the idea of making a convolutional neural network trained on dataset of CP so it could detect CP so that moderation online could be easier. Not only would it detect and potentially automatically remove the content itself but also could go about doing this for both gore, snuff and other content. This CNN would be free for anyone to use and it would solve the proliferation and distribution of CP all over the net by faster and better moderation.

>>6925 >Hi, former pedo here. Why did you wrote this like an youtuber?

Lainchan just needs a robots.txt for /zzz/ and /r/ likely. Apparently it doesnt already have one. A lot of people that set up bots arent that skilled and probably didnt even set them up to ignore a robots.txt. It would be easy. >>6925 What you describe could be abused by feds to further their mass surveillance and imprisonment of dissenters. Also it would constantly trip false positives if you train it on late-preteens and teenagers too since not like their arent older teens and even into 30 something year olds with youthful appearances and it sure would fail at picking up gore and snuff differentiating real from movie and tv show images. AI will be the death of the internet and any remaining privacy anyone has.

>>6925 okey that's sick, make it open-source, you might know better what to focus on because of your specific experiences, make sure to cover what >>6927 said, pitch it at recruitment in programming

but yeah it was my first thought too, what you describe could make a strong competitor on the market given your possible expertise in the matter, if you have skills be careful who you sell them to tldrrrrrr

>>6927 >>6925 Big tech already uses this. In fact, apple has hashes of LOTS of CP that they compare every picture on an apple device against. They use hashes, because having a database of literal CP is illegal. People have already pointed out that this same technology can be used to compare against anything that someone deems questionable like guns, drugs, political symbols, etc. Also wtf is a former pedo? Like 14 year old you liked girls your age? if it's anything but that, you are still a pedo. There is also the problem with ML and things like dolls triggering the system, or all the step-incest porn that features very young looking (but legal) actors and actresses that specifically act/look younger than their ages.

>>6931 Hash checks are different because it's pretty static, use of AI is objectively worse. Apple was going to do that but backed out after major backlash and they dont want to mess up their reputation which will lose them a lot of the market share, they ironically actually have one of the better cloud storage these days from it being encrypted on user side now. You have that more so on Google Drive for example and Microsoft also uses something but Im not sure all places where. Also these hash checks are actually from third party companies that store these hash lists unless they developed one themselves. Huge problem is these hash checks have false positives from matching hashes of different images and it's useless since even the slightest change to an image changes it's hash and a lot of stuff on these databases isnt even anything illegal. Ive heard of cases of people having their Google Drive account shut down from uploading doujinshi.