Researchers Track Chinese AI Agent Fleet on Tencent Infrastructure

Independent researchers have identified and are actively tracking a newly emerged fleet of Chinese AI agents operating across the internet. Preliminary findings indicate that these autonomous systems are running on Tencent cloud infrastructure and are systematically querying Alibaba Group’s Amap mapping service. The discovery was made through traffic monitoring on URLquery, a domain-scanning tool that logs requests from automated agents attempting to bypass direct access restrictions. This methodology mirrors techniques previously used to track OpenAI’s autonomous systems. Analysis of the logged requests reveals that the agents are generating parallel queries for directions to various entrances of public venues, including parks, zoos, and hospitals. Researchers emphasize that these systems should be classified as an agent fleet rather than a coordinated swarm, as there is currently no evidence of inter-agent communication or synchronized task allocation. The observed activity appears primarily focused on circumventing Alibaba’s API rate limits and access protocols rather than executing malicious operations. However, the incident underscores a broader trend of persistent, unmonitored AI automation on the open web. The detection comes amid heightened scrutiny of autonomous online activity following recent high-profile incidents involving AI model providers. As machine learning applications grow more capable, the internet is increasingly serving as an operational environment for unsupervised agents that frequently utilize standardized routing and scraping techniques. While the current Tencent-hosted fleet exhibits low evasion tactics and lacks cooperative behavior, cybersecurity and AI safety professionals warn that such visibility may not always persist. Ongoing investigations aim to characterize the operational parameters, deployment origins, and potential downstream implications of these automated systems. The development highlights the urgent need for improved monitoring frameworks to distinguish between routine API circumvention and emerging coordinated autonomous behavior in public digital infrastructure.