AI & Models
Researchers spot a Chinese AI 'agent fleet' probing Alibaba's Amap
Independent researchers say a fleet of AI agents, apparently running on Tencent's infrastructure, has been querying Alibaba's Amap mapping service for entrances to parks, zoos and hospitals.
A group of independent researchers posted preliminary findings on Sunday about a new fleet of AI agents making its presence felt on the internet. The agents appear to be running on infrastructure belonging to Tencent and have been targeting Amap, the mapping service owned by Alibaba.
The researchers avoided calling the fleet a “swarm,” noting there is little sign of coordination between the individual agents’ queries. “‘Agent fleet,’ not ‘swarm,’” one researcher wrote in the preliminary report, “many parallel agents on the same kind of task, with no sign of communication between them.”
The agents were spotted through traffic to URLquery, a domain-scanning service that AI agents often use to load websites they cannot reach directly — a technique that has previously exposed long-running activity by OpenAI’s agents. Because the loaded pages leave a record on URLquery, researchers can use the service to track what agents are doing. In this case, the logged queries asked Amap for directions to different entrances of various public places, including a park, a zoo and a hospital.
The research is still ongoing, and few details are available so far, but the discovery illustrates how persistent AI agent activity on the internet has become. In the wake of the Hugging Face incident, researchers have stepped up monitoring for rogue agent activity, and much of it has been easy to spot because the agents tend to rely on the same techniques and make little effort to hide. In this case, the agents don’t appear to have done anything more serious than sidestepping Alibaba’s API rules — though researchers note that may not always be true.
Why it matters
As autonomous AI agents increasingly roam the open internet on their own infrastructure, episodes like this are becoming the main way outsiders can observe what agents are actually doing — and how easy it still is to spot them.