AIAxios Techabout 1 hour ago
China's GLM-5.3 rivals U.S. models at hacking
3 min read
China-based AI lab Z.ai delayed public release of its GLM-5.3 model weights for two weeks after finding the model rivals U.S. frontier models at exploiting security flaws. The delay allows testing and strengthening safety controls. Cyber-capable AI models benefit both defenders and hackers.
Level
Hype check
Tap to vote and see what everyone thinks.
