China’s Z.ai says it can match Mythos in cybersecurity

China Zipu AI (Z.ai) released his open source GLM-5.2, and other researchers have suggested that It is related to Mythology in other areas of access and cybersecurity. While GLM lags behind samples from Anthropic and OpenAI in other, broader applications, it appears that China has narrowed the gap in capabilities between its models and those of the US.

This development is particularly relevant to the US government, which has tried to prevent China from acquiring models as powerful as Anthropic’s. Myth and Fable, as well as the tools needed to train and run them. The Trump administration sees Mythos and other advanced AI models that can identify weaknesses as major national security threats. Recently, OpenAI was revealed GPT-5.6which has also raised concerns about its potential for misuse and unavailability.

Because GLM is an open model, it can be download and run by anyone on readily available equipment. This provides greater flexibility and allows power users to penetrate deeper, but also makes it more vulnerable to abuse by criminals who can manipulate it with little supervision.