- Zhipu AI (Z.ai) has released its open-weight GLM-5.2, reportedly matching Mythos in cybersecurity and bug-finding capabilities.
- While GLM-5.2 lags behind models from Anthropic and OpenAI in general tasks, it demonstrates significant progress in specific cybersecurity applications.
- The open-weight nature of GLM-5.2 allows broad access to the model, raising concerns about potential misuse by malicious actors.
China’s Z.ai Narrows AI Cybersecurity Gap With New GLM-5.2 Model
China's Zhipu AI has released its GLM-5.2 model, claiming it can match advanced AI like Mythos in cybersecurity, significantly narrowing the gap with US-developed models despite concerns about its open-weight nature.