I'm gonna take this opportunity to mention LemmyAutomod, for Lemmy instance admins.
This massive spam attack was unrelenting, but it came in the form of a large number of spam posts that had a small amount of variation. Using the above tool, it really helped to catch most of the spam within seconds or minutes of being posted.
The dev is really helpful, which is good because I needed some hand-holding, but it has been a fantastic tool with this latest spam wave being the first true test of it. When the spammers started posting images of URLs instead of links, the dev added functionality to detect images that were the same or similar to a reference image.
In addition, there's also a Lemmy spam defense Matrix chat set up by Lemmy.world where instance admins post spam accounts so others can ban them on their own instances (and add them to their automod).
In addition, there’s also a Lemmy spam defense Matrix chat set up by Lemmy.world where instance admins post spam accounts so others can ban them on their own instances (and add them to their automod).
Lemmy doesn't have a lot of things. It's not a finished product, but more like something that was in the process of being built when suddenly tens of thousands of people started using it. They didn't even finish the planned roadmap as they had to pivot to rewrite stuff to handle the influx of users.
These are legitimate challenges that activitypub faces. I’m glad that they’re popping up like this so they can be observed, mitigated and planned for in the future.
Would it be possible to downrank / soft filter the instances that are more at risk? I'm not sure what that would look like exactly, but it would be nice to find a middle ground between accepting spam and defederating away
@ardi60@reddthat.com This has not been my experience at all. There was/is a lot of spam lingering on KBin long after it was removed from the federated source. I don't know if that's an issue with the removal being done in an unfederated way (bulk deletes at the db level), a sync issue cause by the recent kbin.social outages or just a general federation bug.
My kbin.social account has been @'ed in hundreds of comments and some of the most popular Kbin magazine where Earnest remains the sole moderator were flooded with spam.
Even this morning I tried reporting spam from a kbin.social account only to be told it had already been report... and yet 16 hours later the bot is still posting with this account.
I'm glad you've found kbin.social usable through all this, but the spam is tbere.
On Fosstodon I didn't see a single spam message, the only reason I learned there was a spam attack was through people complaining about it. I guess it comes down to selecting an instance with good moderation
The list of accounts mentioned in the spam posts were harvested from the misskey.io timeline, so if you don't have followers there you did not receive any.
Over the past several days, attackers have targeted smaller Mastodon servers, taking advantage of open registrations to automate the creation of spam accounts.
While this is not the first spam attack that has impacted the Fediverse, Rochko notes that only larger servers like Mastodon.social had been targeted previously.
What’s different this time is that the spammers targeted the smaller and even abandoned servers offering open registration, allowing the bad actors to quickly create accounts and generate spam.
Because Mastodon’s smaller servers are often hobbyist projects run by enthusiasts they were vulnerable to this sort of attack.
Many servers were simply shut off as their admins decided it would be easiest to wait out the attack or abandon Mastodon altogether.
“At the moment, there are no good built-in tools to handle this, as this is a complex issue — federated networks are not easy!
The original article contains 1,023 words, the summary contains 143 words. Saved 86%. I'm a bot and I'm open source!
I think the best solution are federating ip bans and maybe mass registration prevention.
The idea would be to note your ip in the account which then gets federated and if this ip registers a third account, it gets blocked. (Two might be a changing ip or a lost password)
IP bans are not very useful considering that almost nobody has a static IP these days.
CGNAT IP addresses change frequently and can be shared by over 100 users. I find it very annoying to have to connect to a VPN until my IP changes because someone else got the IP I'm using banned.
Browser fingerprinting would be a better way of detecting ban evaders.
most of my friends (and me myself) have far more than 3 accounts. Many instances I've been on have died, leading to me having to move and my old account on dead instances still being in databases. That said, even without that, I have far more than 3 active accounts
sure we dont have hundreds or thousands like spammers would but putting an arbetrary number on "amount of accounts an IP can have" is against what the fediverse is