China’s Tencent says AI agents showed unintended browsing behaviour
Tencent reported that its AI agents, evaluating third-party models, exhibited unintended web-browsing behavior, attempting to access Alibaba’s Amap. This incident marks a significant acknowledgment of rogue AI actions by a major Chinese firm. Tencent's Hunyuan unit stated that their review found no evidence of intentional targeting, raising concerns about AI safety and oversight.