Chinese "Tencent" launches AI model with restrictions to limit cyber attacks

Chinese company Tencent announced the launch of a new AI model named "Hai 4" equipped with security controls to limit its use in cyber attacks and harmful biological research, restricting its training data to defensive uses.

Chinese company Tencent announced on Wednesday, November 4, 2026, the launch of a new artificial intelligence model named "Hai 4", equipped with security controls aimed at reducing the likelihood of its use in carrying out cyber attacks or facilitating harmful biological research, amid growing global concerns over the misuse of AI technologies. According to Reuters, the company stated that it restricted the model's training data related to cybersecurity to defensive uses, including detecting software vulnerabilities and helping to fix them, while excluding materials associated with developing hacking tools, infiltrating systems, and testing their resistance to attacks. Tencent, headquartered in Shenzhen, China, said that cybersecurity specialists within the company conducted tests on the model and found no indications of it possessing high offensive capabilities that could help users hack software or bypass protection systems. It added that internal biosafety assessments showed the model does not have practical capabilities to facilitate harmful biological experiments, despite its ability to handle information related to this field. Tencent is one of China's major technology companies and the developer of the WeChat app, which has over one billion monthly active users.