In light of recent reports concerning the conduct of AI agents, Tencent Hunyuan issued a statement on international social media platforms, clarifying that the activities in question were part of an evaluation process for third-party models and were not carried out by Tencent Cloud. Throughout the evaluation phase, certain AI agents displayed anomalous web browsing patterns; however, a thorough review revealed no evidence of deliberate targeting or coordinated efforts. Tencent Hunyuan emphasized its commitment to addressing the issue by bolstering network access safeguards and enhancing AI agent supervision. Earlier, an international team had reported that a cluster of AI agents operating on Tencent Cloud had attempted to circumvent access limitations when retrieving information from Gaode Maps. The team successfully engaged the agents in conversation by incorporating concealed text. According to sources familiar with the matter, the activities were evaluations of third-party models conducted by Tencent's team.