Open-Weight AI Models Are Catching Up to the Frontier — The Safety Gap Remains

A new SaferAI report says Z.ai's open-weight GLM-5.2 is nearing frontier-model capability while still lacking key safety safeguards.