A fresh wave of AI agents has begun leaving its mark across the internet. Preliminary findings posted on Sunday by a group of independent researchers describe agents that appear to be operating on Tencent's infrastructure while directing their attention toward Amap, Alibaba's mapping service. The researchers were careful to avoid the label "swarm," pointing out that there appears to be minimal coordination among the various queries being made.

"'Agent fleet,' not 'swarm:'" one researcher noted in the preliminary report, "many parallel agents on the same kind of task, with no sign of communication between them."

The discovery came about through monitoring traffic to URLquery, a domain-scanning service — an approach that had previously exposed long-running activity by OpenAI agents. AI agents frequently rely on URLquery to load websites they cannot reach directly, which leaves behind a trail of their activities that researchers can study. In this instance, that trail revealed queries to Alibaba's Amap service, requesting directions to different entrances of various public locations, among them a park, a zoo, and a hospital.

Because the research is still in progress, only limited details are available. Even so, the behavior illustrates just how persistent AI agent activity has become online. Following the Hugging Face incident, a growing number of researchers have been actively watching for rogue agent activity across the internet. Much of that activity has proven easy to spot, since agents tend to reuse the same techniques and make little attempt to hide themselves. In this particular case, the agents appear to have done nothing more sinister than work around Alibaba's API rules — though we may not always be so fortunate.