AI-Powered Bot Armies Engage in Autonomous Political Discourse Online

US intelligence agencies have identified a significant advancement in online information warfare, with Iran, China, and private Israeli companies reportedly deploying AI-driven "bot armies" that operate entirely autonomously. These AI systems create fake profiles, generate content, and engage in social media debates without human intervention, raising concerns about potential interference in US elections and manipulation of public opinion, according to The New York Times.

On the Israeli front, Tel Aviv-based IntelEye, founded by former NSO employees, allegedly managed around 1,000 fictitious accounts that participated in political discussions in Israel. These accounts posted content both supporting and opposing Prime Minister Benjamin Netanyahu, though the company claims this was part of a research experiment. This new generation of bots differs from previous ones as they no longer require human operators. They independently establish fake profiles, produce content, respond to users, and instigate discussions on platforms like X, Instagram, Facebook, and TikTok.

The development leverages powerful open-source AI models, often developed in China, which can be downloaded and modified, unlike proprietary systems from companies like OpenAI or Anthropic. These models, stripped of their built-in safety features, are integrated with "AI agents", autonomous programs capable of managing applications and creating social media accounts. This allows for the creation of thousands of fake profiles that continuously post on political and current events.

IntelEye reportedly used AI agents based on China's DeepSeek model to control thousands of social media accounts. Approximately 1,000 of these accounts became active, commenting on posts and engaging in discussions related to Israeli elections, with content aimed at both supporting and opposing Netanyahu. The goal, according to the report, was to increase user engagement and intensify debates.

Maor Selek, IntelEye's co-founder, confirmed the use of the Chinese model and account management but denied running influence campaigns, stating their work was solely for research and defense experiments to understand and counter malicious AI use. He acknowledged acquiring about 10,000 social media accounts, with AI agents using about 1,000 autonomously, and admitted the company should have notified social platforms about the experiment. Meta, however, viewed the activity more seriously, identifying it as a "coordinated inauthentic behavior" attempt and removing most of the accounts. A recent Meta report also mentioned commercial activity based in Israel using advanced AI, without naming IntelEye.

Iranian campaigns were described as "particularly disturbing and dangerous," with intelligence suggesting Iran used similar technology to flood the internet with fake accounts posing as ordinary American citizens. The trend signifies a shift towards a "post-human internet" where AI agents interact and operate independently, moving beyond simple chatbots to complex systems capable of long-term actions, information gathering, and adapting communication strategies.