Double links for spam

Hi, I’m nephele a developer of Haiku.

I moderate the Haiku forums aswell, and recently there have been spammers signing up an then posting double links, and hiding those as if it were one link with a hyphen. Seeing as discourse already has the “first post very fast” warning, I thought perhaps there could be a warning/flag for new users that include two links, or include links that do not match their actual URI, while having something that looks like an URI in the link title.

This is a screenshot of me removing the spam link, this has happened three or four times now. (each time with the spammer creating a new account)


(I had a screenshot of the second “nice” renderer diff here aswell, but am not allowed to post it as a new user )

3 Likes

I kind of like this idea, though if people want to abuse the system with misleading links or similar stuff, they’ll always achieve this somehow.

I get you’re specifically asking for a feature, but preventing trust level 0 users from posting links could suffice for most cases. Reaching TL1 is fast if you’re a legit user, but spammers never will. So I’m just mentioning the Post links allowed groups setting in case it would help people who come by on this topic.

This is frustrating, but as a TL0 I bellieve you can just post another message with the 2nd picture.

2 Likes

Ah okay, This can probably solve it aswell. Having two links render as one is a bit annoying regardless.

I’ll forward this advice, thanks!

1 Like

Instead of prohibiting TL0 from posting links, you could also reduce the Newuser max links site setting to 1. This could prevent the 2-link problem while allowing the sharing of one link.

New user limitations can be a little annoying for new members, as you noticed when you wanted to post more images.

This is true, but even those who manage a Discourse forum usually complain instead of spending a few minutes reading.

4 Likes

That doesn’t prevent pasting links like https://discourse.org, though, as it count as 1 link.

[https://discourse.org](http://stackoverflow.com)

True, we’ve all gone through this mild inconvenience eh? :smile:

3 Likes

Interestingly this could be something to use AI for. If using:

[Deceptive link](True.link)

Have AI compare links posted this way to see if they match each other.

As it is not using a word description with a link embed. It is using a false link to send someone elsewhere to possible scam.

AI might even be able to identify the double linking the Op mentions as well?

The scan could likely be tailored to also use trust level like tl0-tk1 and be part of spam AI plugin?