Tencent rolls out AI model trained with limited cybersecurity data amid hacking fears
Economic Times
Tencent released an open-source AI model named Hy4, trained exclusively on limited defensive cybersecurity data to prevent potential cyberattack risks.
Chinese technology firm Tencent launched an open-source artificial intelligence model called Hy4 on Wednesday. The company stated that the model was trained using restricted cybersecurity data focusing strictly on defensive content to minimize the danger of aiding cyberattacks.
Tencent filtered out sensitive training materials, including reverse engineering, penetration testing, and exploit development, to prevent the model from acquiring offensive capabilities. This rollout occurs amid growing international concerns regarding AI-driven cyberattacks and threats to global internet infrastructure.
Tencent emphasized its commitment to responsible AI development, stating that it recognizes safety risks related to advanced cyber capabilities. The move follows similar actions by other Chinese firms, such as Moonshot AI, which also limited its newest model's offensive capabilities to mitigate loss-of-control risks.
Why it matters: The release reflects a growing industry trend among Chinese and US artificial intelligence firms to restrict the offensive cyber capabilities of advanced models in order to prevent security risks.







