AI governance researcher at Concordia AI, a Beijing-based consultancy focused on AI safety, said, 'To the best of my knowledge, this is the first time a Chinese lab is ?publicly justifying a delayed open release of model weights with safety considerations. This shows that open-weight risk management practices in China are becoming more sophisticated.' The brand said it had added multiple layers of protection to GLM 5.3, comprising systems to screen risky requests, monitor the model's work and train it to reject malicious tasks.