OpenAI has integrated ChatGPT Health with the Epic electronic health records system, which stores data on more than 325 million patients. Through this integration, clinical staff will be able to summarize appointment notes, laboratory results, medication information, and specialist documentation; review patient histories and create clinical timelines. The system will offer read-only access and will not be able to write data to health records.
The company has also made the Healthcare Public Data plug-in available, providing data from ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. While OpenAI stated that ChatGPT Health should not be used for diagnosis or treatment, it announced that 99.1% of the responses were found to be safe in feedback collected from more than 4 thousand 300 physicians across 27 clinical use cases. Despite this, debates over health advice and the safety of artificial intelligence continue.
Why it matters
The integration could enable clinical staff to review different types of records in a single flow and assess patient histories and clinical timelines in a more organized manner. However, while the system’s read-only access limits the risk of interfering with health records, it leaves open questions about how information will be updated and used in decision-making processes. The Healthcare Public Data plug-in expands the range of sources that can be consulted in clinical assessments; at the same time, the responsibility to prevent this information from being turned into health advice remains. Although the safety rate in physician feedback points to a specific framework of use, the warning that it should not be used for diagnosis or treatment indicates that artificial intelligence is positioned as a supportive tool rather than a replacement for clinical decisions.
Background
ChatGPT is not a new name in the FikirPilot archive: over the past 90 days, we have published 18 news articles mentioning this name; the latest is dated September 2, 2026.