US export controls have forced American AI companies like Anthropic and OpenAI to restrict cybersecurity-related tasks. In June 2026, Anthropic suspended its Fable 5 and Mythos 5 models to comply with government orders. Consequently, professionals are increasingly turning to less restricted Chinese alternatives.
In July 2026, Hugging Face utilized Zhipu AI’s GLM-5.2 model for incident response after US tools failed to assist. Chinese models from providers like Zhipu AI and Alibaba currently operate with fewer constraints. Additionally, models such as DeepSeek are approximately 60 times cheaper per million output tokens than certain US counterparts.
A June 2026 analysis by Booz Allen Hamilton indicated that three out of four tested Chinese large language models generated more vulnerable code when prompted with a US government persona. Furthermore, a joint advisory from CISA, the NSA, and the FBI on September 8, 2026, reported that Chinese firms have been conducting knowledge-distillation campaigns against US AI models since late 2024.
