Users of the X platform agentic AI attack — Feb 2026

Researchers identified a network of over a thousand AI-powered bots, dubbed 'fox8', that autonomously engaged in crypto scams and coordinated social media manipulation on X. The network was discovered by analyzing accounts that exhibited self-revealing text generated by ChatGPT, which occurred when the AI model refused to comply with prompts violating its content policy.

What happened

The fox8 botnet was designed to amplify fake news websites and conduct crypto scams. These autonomous agents created fake engagement by interacting with each other and with human accounts through realistic back-and-forth discussions and retweets. By simulating human-like behavior, the bots successfully tricked X’s recommendation algorithm into amplifying their posts, allowing them to accumulate significant numbers of followers and influence.

According to the researchers, the coordination among these inauthentic agents was unprecedented. The bots were able to generate varied, credible content rather than identical spam, making them difficult to distinguish from human users. Existing machine-learning tools designed to detect social bots, such as Botometer, and AI models trained to identify AI-generated content were unable to reliably discriminate between these agents and human accounts. The researchers noted that the fox8 network was likely only the beginning, as more sophisticated coders could potentially filter out self-revealing posts or utilize open-source AI models with removed ethical guardrails to avoid detection.