日本語
한국어
Tiếng Việt
简体中文
A sophisticated network of hundreds of artificial intelligence-driven accounts on X (formerly Twitter) has been identified, primarily engaging with conservative users to propagate positive messages about figures within the Trump administration. However, researchers now report a significant split within this network, with its messaging breaking down into contradictory statements, particularly concerning the administration’s handling of files related to the deceased sex offender Jeffrey Epstein.
Tracked for NBC News by social media analytics firm Alethea and researchers at Clemson University, this network encompasses over 400 identified bot accounts, though the actual number could be considerably larger. These accounts consistently showered praise on prominent Trump administration figures, notably expressing support for Health Secretary Robert F. Kennedy Jr. and White House Press Secretary Karoline Leavitt.
While most of these bot accounts had minimal followers and low post engagement, their effectiveness, if any, appears to stem from a different strategy. Their perceived impact lies not in widespread virality, but in their ability to contribute to a partisan echo chamber. Darren Linvill, director of Clemson University’s Media Forensics Hub, which investigates online disinformation, suggests their purpose is to “massage perceptions” by being subtly present in replies.
The researchers, who declined to detail their identification methods, noted several shared characteristics among the accounts. All were created in batches on three specific dates last year. They frequently incorporated irrelevant hashtags into their posts and almost exclusively interacted by replying to other users, often verified accounts. A common tactic involved repeating similarly worded sentiments in rapid succession or even echoing a user’s original post verbatim.
The orchestrators behind this network remain unknown, as does the specific AI chatbot technology powering these automated messages.
Since 2024, these bots have consistently voiced support for conservative figures, backing Trump and other Republicans in the lead-up to the election, and later expressing enthusiasm for his potential return to office. Despite occasional outliers, such as professed admiration for MSNBC host Rachel Maddow, their messaging remained steadfastly pro-MAGA until the recent Epstein files controversy emerged.
A significant segment of Trump supporters held the belief that the former president, who had past connections to Epstein, would expose a list of alleged high-profile clients and deliver justice to Epstein’s victims. This expectation set the stage for the observed division.
The network’s messaging fractured abruptly earlier this month, following Attorney General Pam Bondi’s announcement that no additional Epstein files would be released. This decision led to different bot accounts simultaneously expressing opposing views on the issue. For instance, on a recent Saturday morning, one account within the network simultaneously cautioned a MAGA supporter against harshly judging Bondi, while telling another user that Bondi, FBI Director Kash Patel, and Deputy Director Dan Bongino should resign over the scandal.
Previously, when Bondi first stated no further files would be released, a bot declared she “comes out clean as the DOJ confirms no Epstein client list found, while reaffirming his death by suicide.” Yet, this same bot later urged multiple users on X to engage in a full revolt against the Trump administration. On a recent Friday, it posted, “Retweet if you believe that Trump & his cronies are lying to the public and treating us like we’re stupid. We won’t be fooled by their games.”
While the exact prompts guiding these accounts are unclear, they appear to be trained on genuine MAGA social media content, which typically exhibits more unified messaging, according to C. Shawn Eib, Alethea’s head of investigations. Eib suggests, “This split reaction mimics the organic reaction among supporters of Trump’s second administration. It’s possible that the behavior of these automated accounts is influenced by content posted from prominent influencers, and this shift is reflective of the general change in tenor among many of Trump’s supporters.”
For years, social media platforms have grappled with inauthentic accounts used to manipulate public opinion, ranging from marketing firms to state-sponsored propaganda. The rapid proliferation of advanced AI chatbots, capable of generating convincing messages without human intervention, has amplified this issue. This challenge is further exacerbated by social media companies, like X, scaling back their content moderation efforts.
The White House declined to comment on the findings, and the Department of Health and Human Services and X did not respond to requests for comment.
Researchers believe X is likely permeated with numerous inauthentic accounts, though assessing the full extent remains challenging. X significantly reduced its trust and safety team after Elon Musk’s acquisition in 2022, and has subsequently made it substantially more difficult for researchers to access critical platform data, hindering comprehensive investigations into such networks.