Here's a look at a smallish astroturf network targeting @SafeguardDefend (a non-profit group that has released multiple reports on human rights abuses in China) with spammy duplicate tweets. #SaturdaySpam
This network consists of 25 Twitter accounts created in multiple batches in September and October 2022. Thus far, all their tweets were tweeted via the Twitter Web App, and they generally go silent for two days a week (Friday and Saturday in US Pacific Time).
The content tweeted by this network is repetitive, with many tweets duplicated verbatim numerous times across multiple accounts. The most frequently repeated tweets all begin with the hashtag #SafeguardDefenders, and are all attacks on the Safeguard Defenders group.
The accounts in this network use a variety of profile images: real people, abstract backgrounds, cats, horses, and an anime face. Unsurprisingly, the photographs of real people (used by @AdelaideEisenh7, @JaneDav41198277, and @JoyceJe00102569) all appear to be plagiarized.
The cat and horse profile pics show signs of being artificially generated. Note the anomalies in the ears of several of the cat images, the sea of fur surrounding the head in @KristyC92103660's image, and the multiple heads on @BrittneyMangum2's horse (among other things).
Meet @patfkauffman, a "TRUMP 2024!" account motivated by freedom and hard work with over 44 thousand followers. Although this account may at first glance appear authentic, there are multiple problems.
First, @patfkauffman's profile pic is a GAN-generated face. The image has been cropped, preventing it from being detected by facial feature position, but other anomalies remain, such an incorrectly shaped ear and weird artifacts where the hair, skin, and clothing meet.
(More on GAN-generated faces and their use on Twitter and elsewhere here:
As it turns out, the @milanvanacker22 account is one of a set of a dozen cryptocurrency/NFT promoter accounts currently up for sale for impressive prices. The total sale price of all twelve accounts is almost $30,000. #ExtremelyBoringWaysToWasteMoney
These twelve "NFT promoter" accounts (along with a few others) are being sold by a group of four users on accs-market(dot)com: "NFT Promoter", "xGo Agency", "Mark crypto", and the rather boringly-named "accounts".
Unsurprisingly, several (and possibly all) of these accounts have been renamed:
It's a Sunday, and some chucklehead has decided to bestow ~24,000 newly-created fake followers on my account. All are accounts created in November 2022 with 0 tweets, a default profile image, or both.
In a possibly related development, this recent tweet linking a Substack post about detecting fake followers has been spammed with likes from random NFT accounts:
Roughly 40% of my unwanted fake followers also follow @bi3followers1 (ID 1298538907945050112), an account that sells "social media services" and appears to have almost no real followers. This account (which likely violates Twitter's spam policy) has an $8 Twitter Blue checkmark.
Despite being at least seven years old and having few or no tweets, these accounts have somehow accumulated over 50,000 followers each. In an interesting twist, almost all of those followers were created in November 2022.
These newly-minted followers are part of a fake follower network consisting of (at least) 157,342 accounts created between November 3rd and November 13th, 2022. All follow ~100 accounts and have 0 tweets, 0 likes, and few or no followers of their own.
Many of the accounts the network follows are old accounts (created 2015 or earlier) with very few tweets and 50,000 or more followers, all or almost all of which are fake followers from the network. Many but not all are crypto/NFT-themed.
Unmodified GAN-generated faces (so far) have the telltale trait that the main facial features (especially the eyes) are in the same position on each image. This becomes evident when one blends the images together.
There are other signs these images are inauthentic: nonsensical clothing (especially the bizarre "hat" in @JaredPretorious's profile pic), surreal backgrounds, and mismatched ears, among others.
Fake journalist accounts have been a problem on Twitter for a while now, and giving their operators the ability to slap a blue checkmark on them for $8 will in no way improve this situation.
Lots of folks got fooled by @SherylLewellen, a fake journalist account with a GAN-generated face created by far-right group Patriot Front. Does anyone really think groups like this wouldn't drop 8 bucks to make their bogus creations more convincing?
Fake Ukrainian journalist accounts have been a common thing since Russia invaded Ukraine. One, a renamed account masquerading as a journalism student in Kyiv, even got @TheTimes to publish a guest piece. This kind of con would be easier with a blue check.