A network of artificial intelligence-powered propaganda bots promoting Trump administration officials has fractured amid internal MAGA movement divisions over the handling of Jeffrey Epstein-related files, exposing how AI-driven disinformation campaigns manipulate online political discourse. Researchers from Clemson University and analytics firm Alethea identified over 400 confirmed bot accounts on X (formerly Twitter) that automatically generate replies praising key Trump administration figures, including Health Secretary Robert F. Kennedy Jr. and White House Press Secretary Karoline Leavitt.
The AI bot network, operating since 2024, functioned as a digital amplification system targeting conservative users with consistently positive messaging about Trump administration officials. Created in distinct batches during three specific days last year, the accounts demonstrate identifiable behavioral patterns: excessive hashtag usage (often irrelevant to conversations), exclusive engagement through replies to verified accounts, and frequent verbatim repetition of others’ posts. These accounts typically maintain minimal followings – often just dozens of followers, suggesting their purpose isn’t viral reach but subtle perception management within conservative digital spaces.
The Epstein Files Breakdown
The coordinated messaging collapsed following Attorney General Pam Bondi’s July 2025 announcement that the Justice Department would not release additional documents related to convicted sex offender Jeffrey Epstein. This decision ignited genuine division within the MAGA movement, as many supporters had voted for Trump, believing he would expose Epstein’s powerful connections.
The network’s AI systems, likely trained on authentic MAGA social media content, began generating starkly contradictory statements almost simultaneously. Researchers documented individual accounts posting conflicting positions within the same minute – one urging restraint toward Bondi while another demanded her resignation alongside FBI officials Kash Patel and Dan Bongino. Another account shifted from defending Bondi’s “clean” handling to explicitly calling for revolt against the administration: “Retweet if you believe that Trump & his cronies are lying to the public”.
“This split reaction mimics the organic reaction among supporters of Trump’s second administration,” observed C. Shawn Eib, Alethea’s Head of Investigations. “The behavior of these automated accounts appears influenced by prominent influencers, reflecting a general change in tenor among Trump’s base”. The fracture provided researchers with unprecedented visibility into the network’s artificial nature, as human-operated accounts typically maintain more consistent positioning during political controversies.
Anatomy of a Bot Network
The bots represent a sophisticated evolution in computational propaganda. Unlike earlier generations focused on mass engagement, these accounts specialize in “perception massaging,” embedding supportive messages within reply threads to create artificial consensus and reinforce partisan echo chambers. Darren Linvill, Director of Clemson University’s Media Forensics Hub, explains their subtle effectiveness: “They’re not there to get engagement. They’re there to just be occasionally seen in those replies”.
The Epstein controversy highlights a critical vulnerability in AI-driven disinformation: difficulty navigating authentic intra-movement conflict. While effectively parroting unified messaging, the systems falter when confronted with genuine ideological fractures within the communities they mimic. This technological limitation inadvertently exposed the network’s artificial origins to researchers.
Platform Vulnerability and Political Context
Experts note the bot network thrives amid X’s reduced content moderation. Since Elon Musk’s 2022 acquisition, the platform has disbanded much of its trust and safety team and restricted researcher data access, complicating efforts to gauge disinformation’s full scale. The discovery follows previous findings of AI-driven pro-Trump networks on the platform, suggesting an ongoing disinformation ecosystem exploiting X’s enforcement gaps.
The Epstein files controversy represents more than a technical malfunction; it reflects a genuine and growing schism within Trump’s political base. High-profile supporters, including Rep. Marjorie Taylor Greene, Steve Bannon, and Laura Loomer, have publicly challenged the administration’s handling of the Epstein matter, with Bannon warning it could cost “10% of the MAGA movement”. This organic division created an ideological minefield that the AI systems couldn’t navigate.
“This is a case study in how brittle AI-generated disinformation can be when real-world political complexities emerge,” noted Dr. Evelyn Chen, a computational disinformation researcher at Stanford University who was not involved in the discovery. “These systems are trained to amplify, not critically evaluate. When the source material becomes contradictory, the artificial nature of the amplification becomes visible.”
Broader Implications
The incident underscores how AI-powered disinformation tools, while increasingly sophisticated, remain vulnerable to unexpected shifts in online discourse. As political movements naturally evolve and fracture, bot networks risk exposure through inconsistent messaging. However, researchers caution that such networks will likely adapt, developing more nuanced response generation that can navigate intra-community disagreements without revealing artificial origins.
The discovery also raises urgent questions about social media accountability. With X reducing moderation capabilities and restricting external research access, identifying and countering such networks becomes increasingly difficult. This creates fertile ground for AI-powered perception management campaigns targeting not just U.S. politics but global elections and social issues.
As the 2024 election approaches, the fractured MAGA bot network serves as both a warning about AI’s disinformation potential and a demonstration of its current limitations in navigating authentic human political conflict. The digital propaganda arms race continues, but for now, genuine political discord remains a formidable obstacle for even sophisticated artificial influence campaigns.
Subscribe to my whatsapp channel
Comments are closed.