If noone checks if a user is real, will it die?
how is fediverse protected from ai agents?
a month ago by illusionist to c/privacy
ArsonButCute 15 points a month ago
It isn't, like even a little.
I've got a wordpress instance that I've federated into the fediverse and as soon as I did I went from 2-3 spam comments a week to over 50 a week. I've since set up a honeypot and that has dropped to zero. These probably aren't AI agents but they're certainly bots (humans can't see my honeypot unless they're browsing the raw HTML and afict my audience wouldn't do that)
There is absolutely nothing stopping your from spinning up a Masotodon, Lemmy, or any other AP federated platform and unleashing AI agents all across it other than your own conscience.
slazer2au 11 points a month ago
Its not in any way.
There is nothing stopping someone from running up an instance and bombarding other instances with agents.
Sure you can block the instance but all it takes is a new domain registration and you are back to square 1.
DeckPacker 1 point a month ago
That is true, however registering domains isn't trivial or cheap.
So this wouldn't be something you could do en masse and therefore could be managble for admins to ban
slazer2au 1 point a month ago
Na, its quite cheap to get domains. https://porkbun.com/event/holidaysales
DeckPacker 5 points a month ago
Yeah, the cheapest ones are like 5âŦ per year. But to really spam the fediverse and make it impossible to block you, you'd need thousands of them. And that's not really worth it, it would be much easier to just create accounts on random instances, or spam normal social media.
givesomefucks 1 point a month ago
Only because admins are naive enough to federate with anyone...
There's an instance that's literally one person and their "alters".
To keep up appearances it has/had open registration. But if a real person accidentally signs up, they get banned for "impersonation". But for some reason that let the person into admin only communication channels and ability to vote on things as if they were a real admin providing a real service by hosting an instance....
Except theyre the only user on their instance.
The solution is admins stop assuming everyone is sincere. Because all thpes of trolls exploit that
panda_abyss 6 points a month ago
Mods and reporting.
one_old_coder 3 points a month ago
It's been an hour, are you dead yet?
schnurrito 2 points a month ago
It isn't. All that can be done is ban them when they are discovered or defederate from instances that don't do that.
go to feed...
go to feed...
The Fediverse is made up of it's users, mods and admins. We protect it. As a user I report suspicious content and users. The mods and admins decide and take action. As far as I know there's some automatisms as well, like automod scripts. And some code in some of the platforms themselves.
save