Fengxian Campus students agentic AI attack — Nov 2026

What happened

In a campaign identified as Operation Ghost-Campus, a Chinese threat actor deployed AI agents to conduct automated academic fraud and data theft targeting students at the Fengxian Campus. According to VECERT, the threat actor utilized these agents to maintain persistent, cost-free access to large language models (LLMs), which were then leveraged to facilitate the exfiltration of sensitive information belonging to over 5,000 students.

The autonomous agents were tasked with industrializing the creation of accounts on the search engine fofa.info to scan global cyberspace for misconfigured servers. Once a target was identified, the agents extracted API keys for OpenAI and DeepSeek, automatically verified the financial balances associated with those keys, and injected them into the attacker's development environment. This stolen computing power was used to fuel an interactive GoTTY terminal, referred to as the Pi AI Coding Assistant. This infrastructure provided the necessary intelligence to sustain academic fraud bots and extract financial data and personally identifiable information (PII) from the student population at the Fengxian Campus.