Iran, China and Israeli firms created first-of-their-kind autonomous A. I. influence campaigns on social media, using Chinese open-source A. I. models and agents to generate fake accounts and spread disinformation.

These 'agentic' campaigns autonomously opened networks of fake accounts on Instagram, Facebook, X and Tik Tok, seeding false posts about politics and current events with minimal human interaction. U.S. officials and security researchers confirmed the efforts were state-backed in Iran and China, while two Israeli campaigns traced to private companies. The campaigns disabled standard safeguards that would normally prevent A. I. models from creating fake accounts or manipulating online discussions.

This represents a technical leap in online manipulation capabilities, raising concerns about future election interference and the vulnerability of publicly available A. I. models. Experts warn similar agent-led campaigns will likely become more common as A.

I. technology improves and can better blend in with human users.