Chinese technology company Tencent has released Hy4, a new open-source artificial intelligence (AI) model trained with limited cybersecurity-related data, as concerns grow over the potential use of advanced AI systems to support cyberattacks.
Announced on November 4, 2026, Hy4 has been developed with restrictions on cybersecurity content to limit its ability to assist with offensive cyber activities. Tencent said it filtered out training material related to exploit development, reverse engineering and penetration testing, while focusing on defensive cybersecurity applications.
The launch comes amid growing concerns about AI-driven cyberattacks and the ability of increasingly capable models to identify vulnerabilities, automate malicious activities and threaten digital infrastructure.
Tencent said its approach is intended to reduce the risk of the model being misused while retaining its usefulness for legitimate applications. The company also acknowledged the potential safety risks associated with advanced cyber and biological capabilities.
In a statement, Tencent said it was committed to a more responsible approach to AI model training, open-source development and product releases.
The decision reflects a broader shift among AI developers towards introducing safeguards during model development and deployment. Companies are increasingly examining how the capabilities of advanced models could be misused, particularly in areas such as cybersecurity.
Chinese AI developer Moonshot AI also restricted the cyber-offensive capabilities of its latest model in October, according to Reuters. The move indicates that concerns about the misuse of advanced AI are influencing development practices across the industry.
The issue has gained attention as AI systems become more capable of carrying out complex technical tasks. While such capabilities can support cybersecurity teams in identifying vulnerabilities and strengthening digital defences, they can also potentially be used by malicious actors.
Tencent’s release of Hy4 highlights the growing focus on balancing open-source AI development with security considerations. By limiting exposure to certain cybersecurity training material, the company aims to reduce the possibility of its model being used to facilitate attacks while supporting defensive applications.
The launch adds to the ongoing debate over how AI companies should manage the risks associated with increasingly powerful models, particularly as they make their systems more widely accessible.
 model trained with limited....){kind=link}