Opinion: Autonomous AI "Clusters" May Exacerbate Platform Public Opinion Manipulation Risks

PANews February 20th – According to Decrypt, researchers from multiple universities and research institutions worldwide published a paper in Science warning that future online “influence operations” are evolving from easily identifiable simple botnets/zombie networks to “swarms” composed of numerous autonomous AI agents, which can collaboratively manipulate public opinion with minimal human intervention. The study states that such AI swarms can mimic human behavior, adapt messaging across accounts and platforms, and sustain specific narratives over long periods, making it difficult for current risk control and review systems, which mainly target spam content and single-point violations, to effectively detect them. Scholars point out that platforms currently have significant shortcomings in identity verification, account abuse monitoring, transparency of abnormal collaborative behaviors, and external audits, and relying solely on content review cannot form an effective defense.