The only thing keeping the bot population low here is that there just aren’t enough people here to be worth it yet. If the Fediverse grows they’ll come in greater numbers.
It’s also not as SEO-gameable (since fediverse domains are inherently more fragmented than a large, high-reputation domain for SEO algorithms to rank highly), and doesn’t have an inherent monetization system (unlike platforms like Twitter with their ad payouts), so that’s a couple more things going for us.
The only time I’ve even seen the fediverse pop up in a search was when I was doing some black magic nonsense getting Linux to do what I wanted. Yes it was a Linux community no it didn’t help.
The Fediverse has a lot more safeguards in place, in particular the ability to require a message to register an account, such as my instance requires, weeds out 99% of bots.
We can also defederate from instances that become overwhelmed from bots if they have lax sign-up requirements (already happened a few times), which vastly limits their ability to take hold.
The bigger problem for us, I think, is the fight against bot scrapers. Anubis is keeping them at bay for now, but it will likely be an ongoing cat and mouse game until the AI bubble bursts.
the moderation and comment moderation history is nice but it is sort of hidden. that would be nicer if it was more intuitively inline. Also did you mean “bot scrapers” rather than “not scrapers”. And I have been toying with the idea, what about ai or bot supported pipelines, accept that they are part of the visitors, provide it optimized data, and then restrict ui/ux heavy processes to stricter rate limits per second, etc…sort of like a robots txt v2 or something
It’s less hidden for admins. Whenever we click on a user’s profile, there’s a button that can take us directly to their individual mod-log history, which can help us quickly see if there’s a history of AI or bot spam behavior that was spotted by other admins/mods.
Also did you mean “bot scrapers”
Agh, auto correct got me. Yes, I meant bot scrapers.
As for the robots.txt v2 idea, I’m not sure that would be very effective or popular amongst admins, as I think most would prefer not to give it any data if possible, and even with reduced rates, there would probably still be enough bots and queries to present a significant issue.
I think there have been some experiments with ‘trapping’ bots in a recursive loop that they don’t realize is a loop, and therefore can’t escape, but I’m not sure how effective those have been.
For someone who likes digging around in my Reddit history you’d think you’d be able to find something that was actually relevant to the topic at hand. That was a comment about the uses of agentic coding tools for making custom applications, not about building bots.
This particular user is a special kind of one. Once you make the wrong noises about something they don’t like, they make it their personal mission to start digging for anything even remotely looking like dirt that they can sling your way - as they’re now demonstrating below.
I wonder if there is any inherent defence against slop on Lemmy. I guess if an instance doesn’t prune it’s user base of bots, shills and other slop merchants, it could be black listed by other admins of other instances.
No, but there is on PieFed.social. Mods and admins have functionality to check any post or comment for LLM-generated text. There’s more stuff too.
Generally I share my findings with Lemmy admins so they can ban the account.
I think this is a pretty big threat to the fediverse and social media in general and am taking it very seriously. At the moment the amount of slop is pretty low but we need to be ready for the deluge when it comes.
My instance bans non-consensual AI content and users. Rimu’s tools have made that a lot easier to do, I get an alert when the system detects a possible AI user.
People are naive to think there aren’t also thousands of bots here in the Fediverse.
The only thing keeping the bot population low here is that there just aren’t enough people here to be worth it yet. If the Fediverse grows they’ll come in greater numbers.
It’s also not as SEO-gameable (since fediverse domains are inherently more fragmented than a large, high-reputation domain for SEO algorithms to rank highly), and doesn’t have an inherent monetization system (unlike platforms like Twitter with their ad payouts), so that’s a couple more things going for us.
The only time I’ve even seen the fediverse pop up in a search was when I was doing some black magic nonsense getting Linux to do what I wanted. Yes it was a Linux community no it didn’t help.
deleted by creator
No I brute forced it via shit throwing.
More likely it was a DenverCoder9 situation.
The Fediverse has a lot more safeguards in place, in particular the ability to require a message to register an account, such as my instance requires, weeds out 99% of bots.
We can also defederate from instances that become overwhelmed from bots if they have lax sign-up requirements (already happened a few times), which vastly limits their ability to take hold.
The bigger problem for us, I think, is the fight against bot scrapers. Anubis is keeping them at bay for now, but it will likely be an ongoing cat and mouse game until the AI bubble bursts.
the moderation and comment moderation history is nice but it is sort of hidden. that would be nicer if it was more intuitively inline. Also did you mean “bot scrapers” rather than “not scrapers”. And I have been toying with the idea, what about ai or bot supported pipelines, accept that they are part of the visitors, provide it optimized data, and then restrict ui/ux heavy processes to stricter rate limits per second, etc…sort of like a robots txt v2 or something
It’s less hidden for admins. Whenever we click on a user’s profile, there’s a button that can take us directly to their individual mod-log history, which can help us quickly see if there’s a history of AI or bot spam behavior that was spotted by other admins/mods.
Agh, auto correct got me. Yes, I meant bot scrapers.
As for the robots.txt v2 idea, I’m not sure that would be very effective or popular amongst admins, as I think most would prefer not to give it any data if possible, and even with reduced rates, there would probably still be enough bots and queries to present a significant issue.
I think there have been some experiments with ‘trapping’ bots in a recursive loop that they don’t realize is a loop, and therefore can’t escape, but I’m not sure how effective those have been.
https://danluu.com/diseconomies-scale/
removed by mod
For someone who likes digging around in my Reddit history you’d think you’d be able to find something that was actually relevant to the topic at hand. That was a comment about the uses of agentic coding tools for making custom applications, not about building bots.
Yeah, well, I judge you for participating in all those scat and cuck subreddits.
DISCLAIMER: This is a joke. I haven’t stalked them on reddit. Just being silly and want that to be clear. :)
Stop being a weirdo or go start a tech company so your behaviour is at least expected.
Fuck the anti-AI crowd is getting extreme. So much harassment and doxxing going on by these losers.
And it’s not even relevant, using an LLM to vibecode apps is completely different to using LLMs to make posts on social media. Just blind rage.
This particular user is a special kind of one. Once you make the wrong noises about something they don’t like, they make it their personal mission to start digging for anything even remotely looking like dirt that they can sling your way - as they’re now demonstrating below.
I couldn’t imagine being that bored.
It’s linked in his public profile.
Edit: Speaking of blind rage, I hope you downvoted this accidentally…
I wonder if there is any inherent defence against slop on Lemmy. I guess if an instance doesn’t prune it’s user base of bots, shills and other slop merchants, it could be black listed by other admins of other instances.
No, but there is on PieFed.social. Mods and admins have functionality to check any post or comment for LLM-generated text. There’s more stuff too.
Generally I share my findings with Lemmy admins so they can ban the account.
I think this is a pretty big threat to the fediverse and social media in general and am taking it very seriously. At the moment the amount of slop is pretty low but we need to be ready for the deluge when it comes.
Presumably with about the same tools that gleefully give false positives when checking school assignments.
This may come as a surprise to you so brace yourself.
I am not stupid.
Well, firstly, I’ve seen the Piefed code.
Secondly, I doubt it that you have some magical accurate AI-detection tools, and Digg and Reddit don’t.
Your AI detection tools are really helpful, thanks for all your hard work.
My instance bans non-consensual AI content and users. Rimu’s tools have made that a lot easier to do, I get an alert when the system detects a possible AI user.
VPNs on lemmy = punishable by … 👇that’s a bot.
If there are, theyre not very active
I’ve been noticing a lot more definitely-AI-generated “blogposts” and full-on ads in my Mastodon hashtag feeds.
It’s a fairly new thing.
75% of the accounts I’ve had to ban in the roughly-two-years since I spun up my instance have been AI slopfaucets that popped up in the last 2 weeks.