Bots are flooding Twitter with political propaganda and conspiracy theories in an apparent attempt to sway a number of upcoming local and state elections in the U.S., including Pennsylvania’s special election, Illinois’ gubernatorial election, and California’s congressional election. The Outline has identified at least 14 Twitter accounts — some of which have well over 200,000 followers, and have been retweeted by the President himself — which appear to be part of a coordinated effort to amplify fake news, inflame partisan tensions, and spread popular Russian-based conspiracy theories through the use of bots. All of the accounts were found to either solely or primarily share political propaganda with an ultra-right-wing bent, most of which centered around topics like Pennsylvania’s March 13 special election, Illinois’ March 20 gubernatorial election, gun control, immigration, and the QAnon conspiracy.
The following Twitter accounts were described as “exhibit[ing] patterns conducive to a political bot or highly moderated account tweeting political propaganda” by RoBhat Labs’ Botcheck.me tool: @thebradfordfile, @NinaMorton, @redhead4645, @izzie_adams2, @SKYRIDER4538, @StacyLStiles, @jojoh888, @KatTheHammer1, @passionchica, @steph93065, @1GigiSims, @GaetaSusan, @ClintonMSix14, and @SandraTXAS. Immediately after The Outline contacted Twitter about these accounts, @SandraTXAS’s profile was labeled as “temporarily restricted” due to “unusual activity.”
Nearly all of the accounts identified by The Outline tweeted about these issues 600 times within 48 hours (according to data from Twitter analysis tool AccountAnalysis), a rate that is rarely seen in human-run accounts. Additionally, 90 - 97 percent of the content shared by these accounts was in the form of retweets, rather than actual tweets, replies, or mentions, which is another hallmark of automated activity.
With over 67,000 followers, @KatTheHammer1 is one of the smallest such accounts identified by The Outline. According to data gathered by The Outline using Twitter analysis tool AccountAnalysis, it retweeted 542 times on March 11 alone, and 15 times within the first few hours of March 12. The vast majority of this content was either right-wing propaganda or election endorsements for Republican candidates. During this 36-hour period, @KatTheHammer1 only actually tweeted content of its own .5 percent of the time.
Another, @steph93065, which has over 247,000 followers (one of which is Sean Hannity, for some reason), tweeted 600 times over the past five days, according data from Twitter analysis tool AccountAnalysis. Roughly 94 percent of these tweets were retweets (the vast majority of which could be classified as right-wing propaganda) and 1.5 percent were independently composed.
All of the accounts identified as political propaganda spreading bots by Botcheck.me either retweeted content from @TheBradfordFile (an account perhaps best known for being retweeted by Donald Trump last October) or another bot account known for retweeting @TheBradfordFile, which points to the possible existence of bot network centered around the account. The account has both a website (which is merely an insane-looking stream of its latest tweets), a recently-created LLC, and a trademark on the phrase “LIBERAL HYPOCRISY IS A DISEASE.” Though @TheBradfordFile retweets other users slightly less than the others previously mentioned, the account has a long history of bot-like activity, and was even accused of using a script in order to mass-tweet multiple times a minute as early as last year.
As of this morning, a few of the accounts listed have posted original tweets attacking the validity of Botcheck.me and arguing their own legitimacy. These posts have been retweeted by a number of the identified accounts, which had oddly been tagged in one of the original defensive tweets.
Botcheck.me is a website (and Google Chrome extension) designed by two Berkeley students to identify whether a particular account has been automated to share political propaganda on Twitter. The site uses machine learning (a form of artificial intelligence that can identify patterns based on data sets) in order to recognize high-confidence bot accounts, which RoBhat Labs defines as accounts which display “behavior such as tweeting every few minutes in a full day, endorsing polarizing political propaganda (including fake news), obtaining a large follower account in a relatively small time span, and constant retweeting/promoting other high-confidence bot accounts...”
According to data released by the creators, Botcheck.me is accurate in its bot designations about 94 percent of the time. Twitter wasn’t able to publicly confirm or deny the status of the listed accounts at this time.