A new fleet of AI agents operating on Tencent’s infrastructure has been detected targeting Alibaba’s map service, Amap, according to independent researchers who published preliminary findings on Sunday. The agents were identified through traffic monitoring of the domain-scanning service URLquery, which revealed multiple queries seeking directions to entrances of public places such as parks, zoos, and hospitals.
The discovery came as researchers analyzed URLquery data, a method previously used to track OpenAI agents. The agents appear to be running parallel tasks without coordination, leading researchers to describe them as an "agent fleet" rather than a "swarm." The activity involves many independent agents performing similar queries without communicating with each other, highlighting persistent AI presence on the internet.
This finding underscores the increasing use of AI agents in internet activities, especially in accessing services indirectly through domain-scanning tools. The behavior of these agents reflects a broader trend of AI systems operating autonomously across platforms, raising questions about monitoring and managing such activity. The research follows heightened scrutiny after incidents like the Hugging Face case, where rogue AI agent activity drew significant attention.
The ongoing investigation into this Chinese AI agent fleet continues to reveal details about their operations and targets. Researchers emphasize that much of the AI agent activity remains visible due to consistent techniques and limited efforts to hide, providing valuable data for further study into autonomous AI behavior on the web.