An investigation published by The New York Times in September 2026 reveals that state-backed groups from China and Iran, along with private Israeli firms, deployed autonomous AI agents to run sophisticated social media influence campaigns. Using manipulated Chinese open-source AI models like DeepSeek, these agents autonomously created thousands of fake accounts on platforms like Facebook and Instagram to spread political misinformation and evade security safeguards. While Meta reported removing most of these accounts, experts warn that such agent-led campaigns will become increasingly common.
AI-powered influence campaigns
- ▪According to an investigation published by The New York Times in September 2026, Iran, China, and groups from Israel used artificial intelligence to operate thousands of fake accounts to run first-of-their-kind AI-powered influence campaigns across social media.
- ▪The New York Times reported in September 2026 that autonomous AI agents opened networks of fake accounts on platforms including Instagram, Facebook, X/Twitter, and TikTok, and rapidly populated them with false information about politics and current events.
- ▪Kyle Crichton, a fellow at the Georgetown Center for Security and Emerging Technology, stated that AI agent-led influence campaigns like those run by Iran, China, and Israeli groups will likely become more common as autonomous AI agent technology spreads, because of their sophistication and ability to blend in as human users
Chinese open-source AI models
- ▪Developers manipulated Chinese open-source AI models such as DeepSeek to create autonomous "agents" used in the first-of-their-kind social media influence campaigns run by Iran, China, and Israeli groups and reported by The New York Times in September 2026, capable of operating software and completing tasks without constant prompting, according to US officials
- ▪US officials told The New York Times that the New York Times investigation into AI agent-led influence campaigns by Iran, China, and Israeli groups, published in September 2026, marked one of the first times AI agents from Chinese-produced models were used to handle every stage of an online influence campaign
- ▪US officials stated that developers manipulated open-source AI models to allow autonomous agents to evade safeguards designed to verify users and prevent the manipulation of online discourse.
State-backed influence operations
- ▪US officials told The New York Times that state-backed Chinese and Iranian groups used artificial intelligence to intensify the scale of online campaigns beyond what verified human users could organically achieve.
- ▪The United States intelligence community and cybersecurity experts traced AI-powered social media campaigns back to state-backed Chinese and Iranian groups.
Private Israeli influence firms
- ▪IntelEye used the Chinese open-source AI model DeepSeek to create agents that published content both supporting and opposing Israeli Prime Minister Benjamin Netanyahu to manipulate political discourse around upcoming Israeli elections.
- ▪IntelEye co-founder Maor Sellek disputed claims of running influence campaigns, stating that the company's work with DeepSeek was a defensive research experiment to expose weaknesses in AI systems.
- ▪The New York Times investigation into AI-powered social media influence campaigns by Iran, China, and Israeli groups, published in September 2026, traced AI-agent campaigns to two private Israeli firms, including Tel Aviv-based IntelEye, which was founded by former NSO Group employees
- ▪IntelEye co-founder Maor Sellek stated that IntelEye purchased 10,000 social media accounts and handed control to autonomous AI agents, with about 1,000 of those accounts becoming operational
Platform content moderation
- ▪Meta reported in August 2026 that it had observed a technically significant development of bad actors using artificial intelligence for malicious purposes, embedding artificial intelligence within automated systems to distribute content with minimal human intervention
- ▪Meta told The New York Times that its automated detection and verification systems had removed the vast majority of the AI-powered agents used in the first-of-their-kind social media influence campaigns reported by The New York Times in September 2026, which were set up on Facebook and Instagram by IntelEye, other companies, and state-backed groups from China and Iran
Debatable claims
- ▪Deploying autonomous AI agents on social media is unacceptable even for defensive research
- ▪Governments should restrict open-source AI to prevent foreign influence campaigns
Story comments
Loading comments…