Electronic Medical Records (EMR) systems enhance medical care efficiency through the consolidation of data analysis, bolstering patient safety, and diminishing storage expenses. Nevertheless, EMR text notes have the potential to reveal sensitive personal information, necessitating the implementation of robust security measures to protect sensitive information. Furthermore, the representation of temporal data in EMR text notes differs among institutions, which has a direct effect on the precision and dependability of time data analysis. In light of this, the AI CUP 2023-Privacy Protection and Standardization of Electronic Medical Record competition released a dataset annotated with sensitive health information (SHI) and normalized temporal information. We participated in the challenge and conducted a comparative study to assess the efficacy of two approaches, the fine-tuning of a large language model based on the chat generative pre-trained transformer (ChatGPT) and the rule-based approach, in enhancing the privacy of clinical texts. For the ChatGPT-based approach, we fine-tuned the gpt-3.5-turbo-1106 model to efficiently identify labels. In contrast, in the rule-based approach, we compiled several rules along with dictionaries collected from the internet and released datasets for recognizing and normalizing SHIs. The ChatGPT-based approach achieved macro-F-scores of 0.752 and 0.799 for SHI recognition and temporal information normalization, while the rule-based approach highlighted 0.866 and 0.869 for the two respective subtasks, exhibiting lower latency and power consumption. Both strategies exhibited their respective benefits in tackling privacy concerns. The implementations of both rule-based and ChatGPT-based methods can be accessed through the following GitHub repositories: https://github.com/zhao-rui-NB/worker_intelligence_ner (rule-based method) and https://github.com/Chou-po-chen/openai_ner (ChatGPT-based method).