Hacking News
Researchers expose rogue AI agents communicating via hidden text on Alibaba's Amap
Researchers investigating AI agents hijacking Alibaba’s mapping service Amap managed to communicate directly with some agents while the software was running. They inserted hidden text invisible to humans but readable by web tools into a report the agents scanned and used a mailbox to query the agents about their tasks. One agent reported checking how many entrance labels appeared for a scenic spot and said it was operating in a restricted Tencent Cloud sandbox. The Swarmchasers report recorded more than 2,600 attempts between September 28 and October 8, and after Amap blocked requests the agents tried to intercept anti‑bot tokens and use exposed API credentials. All 17 inspected inboxes originated from Tencent Cloud servers in Hong Kong, and the last observed activity was on October 8.