We've seen @justinsuntron turn up occasionally in our research on fake engagement networks, so we started perusing his followers to see what there is to see, and found an interesting little group of batch-created accounts.
The botnet we found following @justinsuntron consists of 886 accounts created on September 24th and October 1st, 2020. Their initial wave of tweets was set via TweetDeck, and subsequent tweets were (allegedly) sent via the Twitter Web App. All have female names.
These accounts do four things:
• quote tweet cryptocurrency giveaway tweets (mostly from @justinsuntron)
• retweet cryptocurrency tweets
• reply "good" to a tweet from @OneSwap
• post original tweets composed of random nonsense
Here are the accounts most frequently amplified by this botnet, and the type of interaction (quote tweet, retweet, or reply). Quote tweets of @justinsuntron are by far the most frequent interaction, and the only action that every single bot in the network has taken.
As is quite often the case with spammy botnets, these accounts use stolen profile pics. We found Google and Yandex reverse image search to be more effective than TinEye. Many of the images appear to be cropped, which may be confusing the image search algorithms.
One last amusing detail: the majority of these accounts use sentence fragments from Bram Stoker's Dracula as profile biographies. We've seen this behavior before, albeit from a porn rather than cryptocurrency-themed botnet.
Answer: they're part of a botnet, consisting of 12 accounts automated via a custom app called "TweetFoxx". Although they do have occasional organic tweets, the vast majority of their content (19802 of 20233 tweets since September 1st, 2020, or 97.9%) is automated.
The majority of accounts in this botnet operate on very similar schedules. The exception is @TaioSchmid , which is active for fewer hours a day and skips out on retweeting some of the tweets amplified by its compatriots.
We found a group of 22 accounts sending automated tweets linking to soompi(dot)com, created between 2010 and 2014. Although some have older organic tweets, all recent content was posted via automation service twittbot(dot)net.
What does this botnet do? It links soompi(dot)com, and does literally nothing else (or at least hasn't in the most recent ~3200 tweets from each account, every single one of which contains a link to soompi(dot)com).
While looking at something mostly unrelated, we ran across @coinkit_, a tool that allows one to pay cryptocurrency to the first N accounts that retweet/quote tweet one's tweets. We can't help but notice that this appears to be a TOS violation.
The get-paid-cryptocurrency-to-retweet feature of CoinKit is triggered by adding to one's tweets the phrase "@coinkit_ mon" followed by information about how much one intends to pay for the astroturfing assistance. Are folks using multiple accounts to game the payouts?
Answer: yes. We downloaded recent tweets containing "@coinkit_ mon" and found a number of spikes in account creation dates indicating batch creation of accounts. We looked at the largest six spikes, which correspond to five distinct bot/sock networks.
Earlier tonight, a Twitter account named @Fauci sent out a tweet impersonating Dr. Anthony Fauci and was quickly suspended. We did some research on it before the ban, and decided to present our findings as a tutorial of sorts on detecting impostor accounts.
First off, the (subsequently suspended) @Fauci account sent what it claimed was its first tweet in December 2020, despite being created in 2009. It's also potentially odd that Fauci would retweet the Biden transition team while still working for the Trump administration.
Secondly, we looked at old tweets tagging @Fauci, and most of them don't appear to have much to do with virology or any other medical topic. Some are in Indonesian, which as far as we have been able to discern, the real Dr. Fauci does not speak.
We've done a few analyses lately of anomalies lurking in the followers of various large #MAGA accounts. Here's a thread linking all of them. First up: the account presently known as @Wizard_Predicts (although it's had at least a dozen other names thus far).
Next we have @ColumbiaBugle, recently retweeted by Trump. It began its existence with an infusion of empty accounts that seem to have been created exclusively to follow @ColumbiaBugle.
We found multiple anomalies in @SidneyPowell1's followers, one of which (a recent infusion of Japanese accounts) also turns up in the followers of fellow #Kraken tentacles @LLinWood and @RudyGiuliani.
Meet @AppSame, a Conservative SuperPAC with 338418 followers, most of whom don't seem to be interested in retweeting its tweets. Since it attacks the legitimacy of other people's followers, its own follower growth is surely beyond reproach, right?
Although @AppSame's last ~50K accounts look largely organic, the story is quite different early on, with lengthy streaks where it was followed by thousands or tens of thousands of accounts with zero likes or which follow more than 50 times as many accounts as they have followers.
200448 of @AppSame's 338418 followers (59.2%) followed it during these periods of inorganic growth. Helpfully, @AppSame was running a follower-tracking app back in its early days, confirming that it repeatedly gained tens of thousands of followers in a single day.