“During a recent security incident, Hugging Face used the open model GLM 5.2 for defense because they were unable to get assistance from a proprietary, closed model provider.”