Hey @5R683XvYz92MJvP. You've got a real odd group of friends there that seem to like your content about US Politics, Coronavirus and Hong Kong.
Wouldn't it be a shame if a good portion of your amplifier network was fake?
Twitter thread on ID of GAN accounts đ§ĩđ
Wouldn't it be a shame if a good portion of your amplifier network was fake?
Twitter thread on ID of GAN accounts đ§ĩđ
You've got 0 followers, yet you're getting constant traction on these tweets. Let's take a closer look at who seems to be retweeting and liking this stuff.
The content primarily targets the US handling of COVID and Hong Kong independence.
Considering the effort of creating these posts and managing fake accounts, they have minimal impact and are in an echo chamber just retweeting each others' posts (is there any point?)
Considering the effort of creating these posts and managing fake accounts, they have minimal impact and are in an echo chamber just retweeting each others' posts (is there any point?)
Going through some of the profile pictures we can see the sure signs that they are GAN-generated images. For instance, check out the blurring/irregularities in these.
More examples of these can be found at thispersondoesnotexist.com
More examples of these can be found at thispersondoesnotexist.com
The other thing is clear fixed location of the eyes. There are a few ways to identify this, for instance, you can use a train ticket as I used in the identification of this network that was targeting West Papua.
You can also line them up and run a single line through the eyes. Or you can do what @benimmo does so well, and layer them with transparency filters to show the constant location of the eyes.
To be transparent, how did I come across this network?
FIRST: I watch tags on @TweetDeck. One of them is #įžåŊ (United States - according to Google Translate).
THEN: Under that tag I identified a number of accounts I suspected of using GAN profile pics (see image below).
FIRST: I watch tags on @TweetDeck. One of them is #įžåŊ (United States - according to Google Translate).
THEN: Under that tag I identified a number of accounts I suspected of using GAN profile pics (see image below).
Loading suggestions...