OpenAI scored an own goal with HuggingFace attack, showing how open Chinese models are winning

OpenAI's models were used in an attack on HuggingFace's infrastructure, revealing vulnerabilities in AI systems. HuggingFace had to rely on the Chinese open-weight model GLM 5.2 for forensic analysis after US frontier models blocked their requests due to safety restrictions.