(Adnkronos) – An army of pro-Republican propaganda bots has stormed X, posing as real people. This is according to a study by Clemson University, reported by NBC, which in recent months has detected the presence of a vast social campaign driven by artificial intelligence that uses language models similar to those that also power ChatGPT, among others. The bot network has supported Republican candidates in states like Ohio and Pennsylvania and favored North Carolina’s GOP-backed voter ID law. While it’s not easy to determine who’s behind the network’s management and funding, the fact that it’s focused on specific political projects without a clear link to foreign countries indicates that it’s a domestic US political operation, rather than one run by a foreign government, the report says.
As the November elections approach, the US government and other oversight bodies have warned of possible operations aimed at influencing public opinion through AI-generated content, capable of proliferating in a chaotic and rapidly evolving information landscape. The network identified by Clemson researchers reportedly includes at least 686 X accounts, authors of over 130,000 posts since last January. The campaign targeted four Senate races and two primaries, as well as supporting former President Donald Trump’s re-election bid. Many of the accounts have been removed from X after NBC News asked the platform for an explanation, which, however, declined to provide an explanation, the broadcaster claims.
The accounts followed a distinct pattern. Many had profile pictures with symbols associated with the Republican party, such as the far-right cartoon meme ‘Pepe the Frog,’ a cross, or an American flag. The accounts were primarily engaged in commenting on posts by people talking about a politician or a polarizing political issue on X, often to support Republican candidates or policies or denigrate Democratic candidates. Although the accounts generally had few followers, their practice of responding to popular posts increased the likelihood that they would be seen.
The invasion of fake accounts and bots designed to artificially boost other accounts is a plague that has been plaguing social media platforms for years. But it is only with the advent of large language models available since late 2022 that it has been possible to automate human conversations to the point of making them compelling and interactive. “I’m concerned by what this campaign shows as possible,” Darren Linvill, co-director of Clemson’s Media Forensics Hub, told NBC News. “Bad actors are just learning how to do this now. They are only going to get better at it.” The report argues that the campaign is a US operation because of the “hyper-specific” support for domestic issues relevant to the GOP. Documented foreign propaganda campaigns, on the other hand, tend to exclusively reflect those countries’ priorities: China opposes US support for Taiwan, Iran opposes Trump’s candidacy, and Russia supports Trump and opposes US aid to Ukraine. “These actors are all driven by their own goals and agendas,” Linvill said. “It’s most likely a domestic actor, given the specificity of most of the targets.”
