OpenAI Launches ChatGPT Health
OpenAI is moving into healthcare with ChatGPT Health, a new product that connects patient portals directly to its AI chatbot. The tool integrates with a user's health data (with consent) to provide tailored information and recommendations. It's a major leap in connecting AI to high-context, personal data, signaling a new frontier for personalized services.
OpenAI's move into healthcare builds on an existing user behavior, with over 230 million people already asking ChatGPT health-related questions weekly. The company says it developed ChatGPT Health over two years, collaborating with more than 260 physicians from 60 countries to review over 600,000 model outputs. For U.S. users, medical record integration is powered by a partnership with b.well Connected Health. This allows users to authorize a connection to their health data from a network of over 2.2 million providers. The system is designed to work with both structured data and unstructured clinical notes from a user's full patient record. Data privacy is a central concern, and OpenAI states that Health conversations are isolated, receive an additional layer of encryption, and are not used to train its foundation models. However, because the tool is a consumer product, the data shared is not protected by the Health Insurance Portability and Accountability Act (HIPAA). The launch of ChatGPT Health was closely followed by Anthropic's Claude for Healthcare, highlighting a key strategic difference in the market. While OpenAI is initially focused on a consumer-first approach, Anthropic's offering is positioned more as a virtual assistant for clinicians to support workflows like automating prior authorizations and research tasks. Medical professionals have expressed a mix of optimism and apprehension. Some see it as a powerful tool to help patients understand their health, while others worry about the "Dr. Google effect" on a larger scale, with potential for patients to misunderstand or misuse the information provided. Concerns also remain about the AI's accuracy and the risk of "hallucinations" when dealing with sensitive medical information. Early independent analysis has raised some safety concerns. A study published in *Nature Medicine* found that ChatGPT Health under-triaged more than half of the medical emergency scenarios presented to it and failed to consistently detect suicidal ideation. OpenAI CEO Sam Altman has publicly stated that users should not expect legal confidentiality for their conversations with ChatGPT, as the legal and policy frameworks for AI are still developing. This has led some privacy advocates to urge caution, suggesting users assume any uploaded information may no longer be private. Initial user feedback has been mixed. Some early users have found it helpful for creating personalized diet and fitness plans based on lab results. However, others have reported that the health-focused model can be overly restrictive and less helpful for more nuanced health questions.
