BuddyPress

Seven spam PM’s. From five different users. In the past 24 hours.

And they’re getting clever too …they’ve found a defect that allows them to delete the backlink to their profile from the message, making it harder to come after them.

I was always shocked by the lack of spam control features on BuddyPress, and I’ve been amazed that nobody has been attacking BP installations …given that with under 1000 lines of code its possible to write a bot that posts billions of spam PM’s, forum posts, and splogs daily to practically any BuddyPress site.

Well, it looks like somebody has written that bot, because the posts are definitely at a volume where its being automated.

If we don’t get this nailed ASAP, we can probably look forward to crippling attacks against pretty much every BP site out there within a matter of weeks.

So I have a proposal: If the core devs give me permission, I can take a few days off working on BP Album+ and write a patch (not a plugin, as it requires mods to the core) that deals with this problem.

@apeatling @r-a-y @DJPaul @21cdb @johnjamesjacoby

What do you think?

^F^

@foxly – Come into the #buddypress-dev irc room on Freenode and let the team know what you have in mind!

You can also use a java web version of IRC if you don’t have a client:

http://java.freenode.net/?channel=buddypress-dev

Initial reaction from Jeff & I is that you detail what you have in mind before you dash off in one direction etc

Okay, per IRC dev chat, let’s use this thread for discussions on ideas to combat registration spam and other types of spam.

Sounds good to me. Give me a day or so to put some thought into it, then I’ll post a more structured proposal.

^F^

And this would specifically deal with spam from actual users who have managed to sign up and and now using the internal messaging system to spam, correct?

And not spam signups, spam blog. Just to clarify.

@Andrea_r

The goal is to limit:

1) Spam PM’s
2) Spam friend requests
3) Spam comments
4) Spam group creation
5) Spam group posts

Once a spammer / troll / hostile has created a member account on the system.

The goal is NOT to stop:

6) Spam comments on blog posts from non-members.
-> Already handled by dozens of plugins

7) Spam in profile fields
-> Limited damage. Will be handled by @francescolaffi ‘s GSoC project

8 ) Spam blog creation
-> Limited damage. Will be handled by @francescolaffi ‘s GSoC project

9) Spam sign-ups
-> Impossibly hard target. The only effective countermeasure is phone verification + geo IP + proxy blacklist; as implemented by Craigslist, eBay, PayPal, Elance, and many others.

Full background on all this stuff in about an hour.

Thanks!

^F^

You rock!

All About BuddyPress Spam

From what I’ve seen over the past few days, the range of knowledge about spam in the BP community ranges from zero to PhD research project. So, to get this thread off to a productive start, I’m going to give everyone some background info on why spammers target our installations, how they do it, and what we can do to reduce or eliminate these kinds of attacks.

1) Why do spammers attack BP communities?

-> Spam is 100% economically motivated. Spammers do what they do because it’s very profitable. Even if only 1 out of a million messages the spammer sends actually reaches somebody, if it cost $2 to send out those million messages and the spammer makes $50 by tricking one person into giving them a credit card number, the spammer is going to throw every resource they have into sending out more messages …because they’re getting a 2500% return on their investment.

-> Given the choice between multiple sites, a spammer will pick the one that gives the largest payout.

Gmail is a “hard” target, with users that are experienced with spam. If a spammer sent a billion spam messages to accounts on Gmail, 99.9% of them would be probably be deleted by automated filters at other ISP’s along the way before even arriving at Gmail. The first thousand messages that arrived at gmail would likely be delivered but would be put in user’s spam folders; and the remaining 999,000 messages would be flat-out refused by Gmail’s servers.

Because anyone with an email account is familiar with spam, probably 999 of those 1000 users would ignore the spam message and 1 user might act on it. So if it cost $20 to send those billion messages and the spammer made $50 by tricking the one person into giving them a credit card number, they’ve only made $30 for all that work.

BP communities are usually “soft” targets that are inexperienced with spam.

Once a spammer gets into a BP community, every single message they send is delivered to a member, and most members are NOT expecting to be attacked by other users on the site.

If a user called “site_news” sends everyone a message that says: “Our site just got featured on Oprah! check out the video! http://www.youtube.com/watch/dQw4w9WgXcQ.cn” every single member is going to get that message, and probably half of them are going to click on the link. (did anyone notice what’s wrong with that “YouTube video” … )

Then, assuming there are 50,000 members on the BP site, half of them click on the link, half of those people are using Internet Explorer, and the attack site the link points to installs a backdoor on computers running IE …at $2 / install the spammer has just made $25,000!

Now, if *you* were a spammer, which site would you attack?

2) How do spammers find BP communities?

Using Google.

Example: http://www.google.ca/search?hl=en&q=%2B“is+proudly+powered+by+WordPress+and+BuddyPress” (front page of every BP site on the net)
Example: http://www.google.ca/search?hl=en&q=inurl:%22/community/members/%22+%2Bbuddypress (members page of every BP site on the net)

3) How do spammers attack websites?

-> Most spam attacks are done using robots, because sheer volume of posts is usually the winning factor. In situations where there is a “captcha wall” or other defense blocking registration to a “high value” site (hint: yours), spammers will use people in low-wage countries to break the captcha and sign up on the site. The going rate is about $2 per 1000 captchas.

http://www.decaptcher.com/client/

Once inside the site, they will then use bots to post spam to all the members on the site.

-> There are literally *thousands* of different programs available that spam websites, and they all have *different* venerabilities.

For example, this program: http://forums.digitalpoint.com/showthread.php?t=1124949

a) Will DEFEAT a “hidden fields” challenge,
b) Will DEFEAT a “javascript proof of work” challenge,
c) Will FAIL a “captcha” challenge
d) Will FAIL an “Akismet” challenge
e) Will FAIL a “Hashed Form Field ID” challenge

But this program: http://www.botmasternet.com/more1/ , wikipedia: http://en.wikipedia.org/wiki/XRumer , video of it running: http://www.youtube.com/watch?v=AL2i4SNPJmg

a) Will DEFEAT a “hidden fields” challenge,
b) Will DEFEAT a “javascript proof of work” challenge,
c) Will DEFEAT a “captcha” challenge
d) Will DEFEAT an “Akismet” challenge (uses proxy networks, never sends the same message twice)
e) Will DEFEAT a “Hashed Form Field ID” challenge
f) Will FAIL a “enter the numbers with a triangle over them” challenge (as used by PlentyOfFish.com)
g) Will FAIL a “click on the photos of cats but not the photos of dogs” challenge

4) How do we stop spammers from attacking BP communities?

-> By making it frustrating and unprofitable (but not necessarily impossible) for spammers to target us; while making these tactics invisible to normal users.

I will cover how I propose to do this in the next post.

^F^

While you’re preparing part 2 I’ll make the comment (probably unpopular) that too an extent this is an issue that BP, WP, Automatic must accept some responsibility for in that WP has always followed the course of making it as easy as possible for inexperienced people to set up a blog/blogs the principle of ‘Out Of The Box’ and ’5 Minute Install’ all designed to promote the app/s to those users who otherwise might be put off, it’s a marketing ploy to ensure that the app gains widespread and popular use (that is being deliberately cynical to make a point) It is due to that that I say there is a duty of care that falls on the App not on the user or community. I know how to get down and dirty with htaccess files, to read logs, enable various methods to deal with an issue – as do many others here – but lets not forget most don’t! I would suggest that it’s time to pull together all the various approaches to dealing with spam in one clear stickied post, make the steps as clear as possible but emphasize that these steps are of paramount importance to follow (thinking about it that may already exist?) Until such time as Foxly or the dev team comes up with the core improvements.

For the record I have enabled most of the steps found in various threads here and elsewhere and also disabled sub blog registration and receive no more than around 6 -8 spam sign ups a day, which we can deal with quite quickly and effectively, I’m still slightly puzzled as to why some appear to have such ongoing issues though, very sympathetic but puzzled nonetheless.

@foxly-

This is a very nice summary of the problem. Thank you for providing the introduction to the various attack vectors spammers currently use.

I would argue, as you know, that WP / BP also needs to combat registration spam–even though it is the hardest issue to address. There area a number of BP.org members that are looking for a solution, however imperfect, that will noticeably reduce spam signups. If a person is infected with a small viral load, the resulting illness often will not be as severe as if they had received a large dose of invading organisms. The same can be said with website spam signups. Any reduction is better than none.

But, as this is your thread and I do not want to take your thread off topic (or have others do what I just did ), I will ask that we table that discussion for another thread at another time and focus in this thread on solutions to combating spam once a spam account has successfully registered.

Once again, this is a great start to the conversation.

Maybe BP should require that you choose your own register slug after activating the plugin. Perhaps also require you name your required fields fields, instead of the default “name” or “base.” The less default settings BP has the harder it is for BOTS.

Allowing users to report spam would be a very useful feature. There’s only so much you can do in terms of technical spam prevention, and technical spam prevention always gets cracked eventually.

If the amount of spam reports for a certain account exceeds x, then freeze the account until an admin can review the account. The admin should then have the option to do an IP based ban of the account if it appears to be a spammer. Some very basic IP based messaging/commenting/posting rate limitation would help too.

My BP site is fairly new. I had one PM spammer and I changed my register slug and added birth day to the required fields and so far no return spammers (about 1000 new members per month, 4,000 current). I’m sure this won’t end attacks, but hopefully it with stave off many of the BOTS.

Example: http://www.google.ca/search?hl=en&q=%2B”is+proudly+powered+by+WordPress+and+BuddyPress”; (front page of every BP site on the net)
Example: http://www.google.ca/search?hl=en&q=inurl:%22/community/members/%22+%2Bbuddypress (members page of every BP site on the net)

Very much behind this, but I will mention that changing those two things are the first thing I’ve done with my BP installs (along with other stuff I mentioned in the article I did for the I-guess-it’s-not-coming-back bp-tricks.org). Agree that an install routine that forces the user to customize their slugs (explaining possibly consequences if they don’t) would be a great idea.