A previously undisclosed network of hundreds of AI-powered accounts on X has been automatically replying to conservatives with positive messages about the Trump administration, according to researchers.
The network’s automated nature was revealed when its messaging fractured over the administration’s handling of files related to deceased sex offender Jeffrey Epstein. The accounts began issuing contradictory statements on the topic, exposing their AI-driven origins.
The network, which was identified for NBC News by social media analytics firm Alethea and researchers at Clemson University, includes more than 400 known bot accounts, though researchers believe the actual number could be much larger. The accounts consistently praise key figures associated with Donald Trump, such as White House press secretary Karoline Leavitt, and have also voiced support for Robert F. Kennedy Jr.
While these accounts typically have few followers and their posts receive minimal engagement, a large audience does not appear to be their objective. Darren Linvill, director of Clemson University’s Media Forensics Hub, which analyzes online disinformation, explained that the network’s goal is to contribute to a partisan echo chamber and “massage perceptions” through the sheer volume of posts. “They’re not really there to get engagement,” Linvill said. “They’re there to just be occasionally seen in those replies.”
Researchers noted that the accounts share several distinct characteristics. They were all created in batches on three specific days last year and almost exclusively post replies to other users, particularly those with paid verification on X. The bots frequently use hashtags irrelevant to the conversation, repeat similarly worded sentiments in quick succession, and at times, simply copy and paste a user’s original post back to them in a reply. The identity of the network’s operators and the specific AI model used to power it remain unclear.
The bots have posted pro-conservative messages throughout 2024, supporting Trump and other Republican candidates in the lead-up to the election. While their messaging was generally consistent, it became erratic following the recent controversy surrounding the Epstein files. A core group of Trump supporters believed he would expose a list of Epstein’s powerful clients, but the AI network faltered when Attorney General Pam Bondi announced she would not release additional files.
The split in the MAGA movement over Bondi’s decision caused the bot network’s programming to break down, with some accounts posting opposing opinions almost simultaneously. In one instance last Saturday morning, an account from the network told one user to avoid judging Bondi too harshly, while telling another user in the same minute that either Bondi, “FBI Director Kash Patel,” or “Deputy Director Dan Bongino” should resign over the matter.
Source link
