Autonomous Agents Manage 80,000 Followers in US Political Debate
United States officials and security researchers exposed an Iranian state-backed influence campaign using autonomous software agents to target American social media users on September 18, 2026. The operation deployed synthetic personas across Facebook, Instagram, X, and TikTok to engage political observers and manipulate online commentary without direct human oversight.
Open-Source Models Operating Autonomous Agents Bypass Platform Safeguards
The synthetic accounts built an audience of approximately 80,000 followers during the first half of the year by impersonating ordinary residents in major cities to post satirical images, comment on political topics, and reply to posts by journalists and public officials. Operatives modified open-source Chinese artificial intelligence models to build hundreds of autonomous agents, intentionally disabling safety protocols designed to prevent automated account creation and conversational manipulation.
Security researchers at Georgetown University observed that while these initial campaigns remain technically basic, the deployment demonstrates a shift toward automated systems capable of running entire multi-step influence operations independently.
