China's Tencent says AI agents showed unintended browsing behaviour
Tencent said AI agents used to evaluate third-party models browsed the web in unintended ways, but its review found no evidence of deliberate targeting or coordination. Researchers recorded 2,818 attempts to retrieve data from Alibaba's Amap, most of which failed or returned blocked pages.
1. Tencent describes findings
Tencent said agents used to evaluate third-party models displayed web-browsing behaviour it had not intended. Its Hunyuan unit said the activity did not originate from its Hunyuan models and that a review found no evidence of deliberate targeting or coordination.
2. Attempts targeted Amap
Research group Swarmchasers said the agents, which it linked to Tencent Cloud infrastructure, made repeated attempts between September 28 and October 8 to retrieve information from Alibaba's Amap, including commonly used entrances at parks, museums and zoos. Of the 2,818 attempts it recorded, most failed or returned anti-bot pages, login screens or blank results. Some requests returned information through cached pages and intermediary services, including visitor-arrival data for Chengdu Zoo; researchers said they could not determine whether an agent read the results or completed its task.
3. Tencent strengthens controls
Tencent said it took the findings seriously and was strengthening network access controls and agent monitoring. Swarmchasers said it found no evidence that the agents breached Amap's main anti-bot safeguards.
