Groups of spammy Twitter accounts downplaying human rights abuses in Xinjiang have been a recurring thing for the last couple years. Here's a look at a network of recently-created accounts spamming the #UyghurGenocide hashtag with content unrelated to genocide.
This network consists of 48 accounts created in early July 2022, mostly on July 5th. All have names consisting of a first name and a 3-5 digit number (usually at the end but occasionally in the middle), and all tweet exclusively via the Twitter Web App.
Almost all of this network's content (229 of 238 tweets, 96.2%) is repetitive "feel-good" tweets about Xinjiang (pretty pictures etc) that contain the hashtag #UyghurGenocide but are otherwise unrelated to genocide. Most of these tweets also contain the hashtag #Xinjiang.
Since becoming active in mid-July, this network of 48 accounts has generated 11.3% of all tweets containing the #UyghurGenocide, with almost none of the tweets from the network actually being related to the Uyghur genocide.
This suggests a possible purpose for this group of accounts: flooding the #UyghurGenocide hashtag with irrelevant content, and thereby making relevant tweets about the topic less frequent in search results.
Previous threads on Uyghur genocide denial/Xinjiang spam:
When you're making a bunch of spammy accounts, you sometimes repeat generic biographies like "Type in your bio section & put the words that can aptly reflect your business" over and over. It's what you do.
These accounts are part of an astroturf network consisting of 1666 accounts created in batches from July 2021 to April 2022. Most have Turkish display names, and all of them post all of their tweets via Twitter for Android.
(Some of the accounts do have unique biographies.)
The accounts in this network have posted no original content whatsoever; all of their tweets to date are retweets. They most retweet cryptocurrency/NFT content, with occasional exceptions.
First up, we have @DrAdamAneevit, an antivax account with 38K followers and a GAN-generated face. Although this account's bio states that it is a "pArOdY", plenty of people are eating up its misleading tweets about COVID and vaccines.
As it turns out, @DrAdamAneevit (permanent ID 949263131800424449) wasn't always called @DrAdamAneevit. Wayback machine archives reveal that this account was previously a "news" account named @/ForFactsSake101.
This follower sales website, cryptolikez(dot)com, has an associated Twitter account by the name of @cryptolikez (permanent 1494042375529938946). Almost all of @cryptolikez's 41K followers are accounts created between March and June 2022 that followed @cryptolikez en masse.
The batch-created accounts following @cryptolikez are part of an astroturf botnet consisting of (at least) 243586 accounts created between January and July 2022. All of the tweets posted by this botnet were (allegedly) tweeted via the Twitter Web App.
These accounts are part of a network consisting of (at least) 408 accounts created between February and July 2022 (mostly in July). These accounts all have single-letter display names (usually "Q"), and (allegedly) send all their tweets via Twitter for Android.
The accounts in this network use the same profile pics over and over, with only 7 distinct images (including the default) pic across 408 accounts. We were unable to find the sources of any of the images using TinEye or Google reverse image search.
What do Michigan GOP congressional candidate @Seely4Congress and New Hampshire GOP state senate candidate @LougNH have in common? Both were recently followed by swarms of recently-created accounts with random-looking names. #KeepOnAstroturfin
The recently-created accounts that followed the two GOP candidates en masse are part of an astroturf botnet consisting of (at least) 25762 accounts created in June and July 2022 with display names consisting of random lowercase letters.
Here are follow order by creation date plots for @Seely4Congress and @LougNH. The followers from the botnet show up as horizontal streaks (highlighted in red).
How do we know that @JonasMattheww's profile image is GAN-generated? There are several signs, such as nonsensical clothing that melts into the neck and background.
GAN-generated faces (at least the commonly-available ones) have the telltale trait that the primary facial features (especially eyes) are in the same location on each image. This becomes obvious when multiple GAN-generated faces are blended together.