“Some Chinese large language models have incorrectly identified their creator as OpenAI, fueling speculation that they were trained on the outputs of American models through a process known as distillation.”