US, international authorities seize Russian AI bot farm
The Department of Justice said Tuesday that it had seized a pair of internet domains and hundreds of social media accounts that authorities claim were part of a Russian-government bot farm that used AI to generate content aimed at meddling in American politics.
According to warrants issued in an Arizona district court late last month, Russian operatives relied on a pair of domains to register 968 unique accounts on the social media platform X. The warrants allege that the accounts were created in 2022 at the behest of the deputy editor-in-chief of the Russian state news outlet RT.
The email addresses and phone numbers used to purchase the domains were traced to a person identified as the lead developer of “Meliorator,” which authorities described as a “covert artificial intelligence-enhanced software package.”
The two warrants — issued June 27 in the U.S. District Court of Arizona — order the domain name registrar Namecheap to redirect the domains in question, “MLRTR.com” and “OTANMAIL.com,” to servers controlled by the FBI.
According to a joint advisory released Tuesday by the United States, the Netherlands and Canada, “affiliates of RT” used Meliorator and its AI capabilities to create a bot farm featuring false personas of various nationalities in order to spread disinformation and narratives favorable to the Russian government on X.
Meliorator can create social media personas “en masse,” post content, mimic the tone of ordinary social media users and align the substance of their posts with other disinformation accounts and bots, according to the advisory. There was also evidence that program developers were working to expand its functionality to other social platforms.
The operation targeted audiences in the United States, Poland, Germany, the Netherlands, Ukraine, Spain and Israel.
Tuesday’s law enforcement operation comes as U.S. intelligence officials and private threat intelligence organizations have warned in recent months that the focus of Russian disinformation operations have shifted to focus more heavily on the upcoming U.S. elections.
Groups traced to the Russian government, like Doppelganger and CopyCop, are increasingly incorporating generative AI tools into their propaganda efforts, though reports by industry players like Meta and OpenAI have cautioned that these efforts have largely been unsuccessful at reaching large audiences and gaining traction online.
“Today’s action demonstrates that the Justice Department and our partners will not tolerate Russian government actors and their agents deploying AI to sow disinformation and fuel division among Americans,” Deputy Attorney General Lisa Monaco said in a statement.
Citing information obtained from an unnamed U.S. agency, the FBI claims that the RT deputy editor-in-chief pitched the outlet’s leadership on the development of the new software prior to 2022, when they were looking for “alternative means for distributing information beyond RT’s standard television news broadcasts.” The network’s RT America channel would later shut down in March 2022 after it was dropped by DirecTV and Dish Network.
Leadership at RT approved the development of Meliorator, and less than a year later the project was folded into a new private intelligence organization created by the FSB and approved by top Kremlin leadership in order to sow discord in the United States, according to court documents.
In response to a request for comment from CyberScoop, Anna Belkina, RT’s deputy editor-in-chief and head of communications, marketing and strategic development, said she’s “more than happy to tend to my farm (dacha) — which is made up mostly of tomatoes and strawberries.”
Bots featured in the Russian campaign responded directly to posts on X from candidates running for federal office in the United States, pretending to be from the candidates’ home district. The posts pushed back on U.S. support for Ukraine and claimed that countries like Poland, Ukraine, and Lithuania owe Russia a debt of gratitude for “liberating” them from Nazi rule during World War II.
An FBI official wrote in an affidavit that seizure of the domains “would have a significant impact on the actor’s malicious activities,” preventing them from using multifactor authentication needed to pass X’s bot filters, stopping the creation of new profiles using emails from the seized domains and blocking delivery of emails to the servers.
