Open-weight AI models are catching up to the frontier. The safety gap remains.

HeadlineFlip summary

A SaferAI report indicates Z.ai's open-weight GLM-5.2 model nears frontier AI capabilities. However, it lacks crucial safety measures, raising concerns about powerful open models outpacing governance and safeguards.