Chinese AI manufacturers lower the network attack capabilities of new models
Alibaba, Tencent, and Dark Moon recently released new models that reduce the capability for cyber attacks and enhance cybersecurity functions. Alibaba filtered training data for Qwen4-Max, retaining cybersecurity knowledge used for detecting and fixing software vulnerabilities while removing materials aimed at enhancing attack capabilities. This model also employs specialized datasets and post-training improvements to reduce risks related to chemical, biological, radiological, and nuclear threats.
Tencent stated that the cybersecurity training for its Hy4 model is limited to defensive purposes; Dark Moon previously reduced the attack capabilities of its model when launching Kimi K3.1 Pro. Alibaba reported that Qwen4-Max scored 14.7% in secure code generation benchmarks, compared to 8% for the previous model; the company also plans to release a small open-weight model in the Qwen4 series.
ChainCatcher reminds readers to view blockchain rationally, enhance risk awareness, and be cautious of various virtual token issuances and speculations. All content on this site is solely market information or related party opinions, and does not constitute any form of investment advice. If you find sensitive information in the content, please click , and we will handle it promptly.






