The outbreak of COVID-19 has led to a global surge of Sinophobia partly because of the spread of misinformation, disinformation, and fake news on China. In this paper, we report on the creation of a novel classifier that detects whether Chinese-language social media posts from Twitter are related to fake news about China. The classifier achieves an F1 score of 0.64 and an accuracy rate of 93%. We provide the final model and a new training dataset with 18,425 tweets for researchers to study fake news in the Chinese language during the COVID-19 pandemic. We also introduce a new dataset generated by our classifier that tracks the dynamics of fake news in the Chinese language during the early pandemic.