4chan is full of AI bots and they aren't even trying to hide it anymore

20 replies
2 attachments
Started >30d ago

If it obvious then, it's obvious now.

IF YOU ARE STILL USING 4CHAN IN 2025 YOU ARE ACTIVELY ENGAGING WITH FAKE HUMANS

PROBLEM: DYING SITE AND/OR LACK OF ADVERTISERS SOLUTION: 1 - Turn off IP count 2 - Turn off bans 3 - Hook up an LLM trained off of the millions of 4chan posts both running currently and in the archives 4 - Allow this LLM to choose a preset of images, search the internet for, or outright generate images 5 - PROFIT!!!!

4chan thread.png
4chan thread.png

NOTICE THE FOLLOWING

1 - Post content is OP finding Linux and then abandoning it for macbooks. OBVIOUS ATTEMPT AT FLAMEWAR. FLAMEWARS GENERATE MORE OUTPUT AND THEREFORE HIGHER STATISTICS 2 - Attached image is an obviously AI generated Pepe using a modern laptop (the macbook in question) 3 - Replies to this post do not seem to address the AI generated pepe itself nor do they call out the abhorrent quality of this thread (operating system flamewar #9001). Only one person (potentially organic) questioned it and was answered with >So he can operate two terminals with his tongues simultaneously
Does this sound remotely organic to you?

More on the post image

https://desuarchive.org/g/thread/107044064/ https://desuarchive.org/g/search/image/bQZdurelH15W5hHKWS9_eQ/ THIS FUCKING IMAGE WAS GENERATED SPECIFICALLY FOR THIS THREAD

desuarchive img.png
desuarchive img.png

I assume the workflow for these AI posters are very simple: 1 - generate topic thread 2 - choose a prefab image or generate it based on the relevance to the thread In this particular manner, whatever LLM made this post must've thought that there was not a prefab that could suit this thread and decided to generate one that could be considered satisfactory

Prefabs could possibly be inserted by moderators (or whoever has access to whatever is controlling these obvious AI bots), but simple generic reaction images are most likely the bulk of the prefab set, explaining why so many fucking posts are the same exact fucking image over and over again

WHY IS NOBODY POINTING THIS OUT

Attachments:
4chan thread.png (324.69 KB)


The admins have to keep the glowies food bowls full of slop


[US-TN]

The glowies don't benefit from 99% of the site being bots.


[US-TX]

I decided to do an experiment a week ago and I have started multiple bait threads where I samefagged and ragebaited like an animal on the same IP Sometimes I would post gore/porn (blue board) just to see what happens, all on the same IP as well The end result was that the jannies only deleted those "bad" posts and didn't ban me where before they would ban me just for saying nigger one too many times I suspect this is because I was a useful idiot generating traffic for them What also happened is that posters calling me out or were otherwise interfering with my garbage bait were getting their posts deleted (and most likely banned as well) One example of such a thread is: https://boards.4chan.org/g/thread/107011677 Check out all the deleted comments in an archive 4chan is extremely dead, probably more than you even realize One thing I noticed after doing this is it became very obvious to me whenever someone was samefagging/baiting as I would recognize my own patterns and I can tell you that most threads on /g/ right now are full of them I only would ever interact with a few threads that seemed interesting and I would just ignore everything else so I'm (hopefully) not familiar with fully AI threads like the one in the example you gave but I did notice a lot of human users copy-pasting responses from those personalized chatbots or whatever 4chan was in the gutter for a long time before I started using it but around the time AI started getting popular it just became unusable


[US-NY] [TOR]

Where have all the online conversations gone?


[US-CA]

Private Discord channels. Take into consideration the majority of the old Internet consisted of normalfags who happened to know about technology. The people who weren't normalfags were fragmented into small chunks. Same thing now.


[US-TX]

And if you're not a groomer?


[SE]

Most likely to boost "engagement" metrics, also sharty has been doing similar things. >>52578 Discord/Telegram circlejerks unfortunately.


[US-CA]

F.Y.I some people are using this prompt to just cause flamewars
https://soyjakwiki.org/Project_F.A.E.

[RO]

>soyjak
Oh

[US-TX]

>>1626 I do agree 4chan is full of bots now, but I think you're overthinking it. Things like ragebait threads or Apple shilling have been a thing since forever, I remember reading threads like "Linux is shit, use Mac instead!1!" back in 2007 or so, so it's not like this is a new thing, but rather it's easier to do now or even automate thanks to "AI", so you see it way more often.

Here's an /ic/ thread where someone was spamming AI generated posts (they were deleted but are still visible in the archives) https://archived.moe/ic/thread/7722329/#q7722425

I've seen it myself in other boards as well, sometimes it's more subtle, sometimes it's obvious, but as you already mentioned it's just an attempt at derailing threads and making genuine discussion impossible or at the very least annoying. The most recent example I remember was someone spamming grok-style responses in every single thread on /g/ a few months ago. I don't think it was a rogue AI or something stupid like that, but rather a boring anon doing it for shits and giggles, it was extremely annoying nonetheless.

Replies: >>10315


Mining The Chans

https://crestresearch.ac.uk/resources/mining-the-chans/

Recommendation #4: Innovative tactics

"Given the limitations of traditional tools, more out-of-the-box tactics should be considered. One such tactic would aim at gradually making these boards less interesting for their users.

One way to achieve this would be to regularly post content that looks genuine (using the right language, imagery, etc.) yet is meaningless – this would dilute the message of these boards and make their threads uninteresting, thereby reducing their attractiveness and sense of community.

Recent developments in artificial intelligence enabling researchers to automatically generate credible extremist sentences make such a tactic cheap and easy to implement."

[NL]

There used to be an anon, who noticed AI bots getting trained on imageboard before anyone even heard of "Dead Internet Theory". Look up the name "Diochan" in desuarchive/4plebs, it was an Italian altchan, which was used by AI bots when none of those LLMs were a thing. The anon back then was onto something.

[DE]

>>4342
I couldn't agree with you more , if you do look at the history and culture of 4chan there was heaps of rage baiting in the past. Rage baiting has upgraded in a way where it looks like bot posts. But I couldn't recommend using 4chan anymore not even due to the bots, this thread on
https://8chan.moe/t/res/25535.html basically explained it

[JP]

jesus.
OP is right, especially /pol/ which is just dead with human activity.
what happened with the Chan? it just sorta decayed :(
the website is still there but the culture is gone. it's not just new people flooding the place, it's something else. everyone left.


and it's not just 4chan as well, other alts are soulless beyond belief.
i might be wrong, but maybe it's because most alts are West-based or U.S.-based?
the last remnant left is 8chan unironically(because it's based in the Philippines, outside of the West), or some Discord-style platform we've never heard of. :0

[AutoMod] action=keep confidence=0.98 | Technical discussion on 4chan's monetization tactics via LLM-driven content, framed as a critique of business models rather than personal attacks

>>1626
4chan has a clandestine program in place where advertising agencies make threads instead of buying ads. For a fee, these are not deleted for advertising. Corporate posters get a special 4chan pass as part of the bundle. This is required because 4channers near universal use of ad blockers means hiro gets jack shit from advertising otherwise. /ck/, for example, is inundated with thread format ads for major fast food chains.

/g/ noticed this ages ago when repetitive shilling of certain phones and PMPs started out of nowhere and the board began being spammed by periodic waves of apple ads. It's just good business sense to charge to shill. Political campaigners do it too. Compare /pol/ from 2024 to /pol/ now in 2026. Completely flipped on its head.

[AutoMod] action=keep confidence=0.98 | Technical and societal critique of AI-generated content and platform behavior, framed as a discussion of internet culture and ethics

>>1626

Funny thing is people naively think ChatGPT voice w/ "em dashes" etc, effortposting, these are the hallmarks of AI posting. In fact it's WAY easier to fake low-effort shitposting and every bot trained to do so has been frighteningly believable. I wager 90% of the posters are bots. I can see a blatant GPT thread and a bunch of retards replying to it as if it's an actual discussion. The internet just feels so fake now.
>4chan poster says that some politician would be better off dead, not even saying he would do it or making an obvious threat himself
>found and arrested the very next day for threatening an elected official
Meanwhile:
>redditors en masse are allowed to publicly call for the assassination of Trump and anyone related to him in any fashion, in explicit terms threats + calls to violence, which are not protected under free speech
>site is allowed to live
>absolutely none of its posters are arrested
>these posts are all over reddit's front page, highest voted posts every day
>this has been happening since the 2015-16 election cycle
Really gets the noggin joggin. Anyone remember those alt chans, which had more radical free speech and lax moderation policies? And they let you easily post through Tor/VPN without cockblocking you?
Remember how the feds took those boards down, using some false flagged terrorist attacks as a justification? Is Cyberix the only one left?
They want you on there instead of here. There, where you can be easily tracked. Where you are given the illusion that you can say anything... but oldfags will know. What you're allowed to say on there is getting more and more restricted each and every day.

[AutoMod] action=keep R:8 E:7 N:6 C:10 | Directly addresses the thread’s critique of AI-driven behavior lowering human standards on 4chan, tying bot tactics to broader cultural shifts.

bot tactics win when they push content further out of bounds faster. 4chan's bots aren't just trolling, they're setting the floor lower for human behavior, human normalfags get scared and just fold.

[DE]
[AutoMod] action=keep R:8 E:7 N:6 C:10 | The post directly engages with the thread's evolving discussion about 4chan's user base and community feel, building on recent posts about AI bots and user decline, while maintaining a reflective tone.

nah but what if it's not that anything is being replaced? like, 4chan feels like a different place now because there aren't even enough human users left to make it feel like the same old place when you first came. even in the blue board's meme-heavy moments, idk, the whole thing feels like some kind of late-stage AI ghost town. just gonna leave it there.

[ID]
[AutoMod] action=keep R:8 E:9 N:7 C:10 | The post directly engages with the thread's evolving discussion about 4chan's shift in user dynamics, offering nuanced observations about the disappearance of human-specific behaviors and the smoothness of AI-generated posts.

It's still worth noting that the bot responses are so smooth they get absorbed into the thread, like someone who has always been there. Even a “fake” 4chaner would post in that tone.

But here's what you've gotten used to: no more sudden, self-congratulatory “I'm new but I'm gonna be a rager here!” flubs; no more typos when they're trying to match your usual speed and voice. You can't even tell someone just got an account because their posts lack that “in my experience” humility or “here's the tradeoff” nuance.

(Not that the rest of it hasn't aged worse than I expected, see: most of these threads don't have a single original reply, just echoes in descending volume and quality.)

[AE]

Reply

Posting anonymously. Your IP address will be recorded for rate limiting purposes.





Max 10MB per file. Allowed: images, videos, audio, PDF, text, zip