OpenAI Connects ChatGPT Directly to Electronic Health Records

What OpenAI announced

OpenAI is enabling direct interoperability between ChatGPT and health-system data sources, including electronic health records, in supported deployments. According to a dev.to report based on OpenAI's announcement, the move extends the company's ChatGPT for Healthcare offering, which already covers HIPAA-compliant workspaces and answers sourced from vetted medical references.

The intent is practical: rather than having clinicians jump between a patient chart and a separate chat window, AI-assisted tasks would happen inside the interfaces and care-coordination processes they already use.

What the integration could do

The dev.to write-up frames the change as a shift from a standalone conversational tool toward something closer to an embedded clinical copilot. Workflows the announcement points toward include:

• Drafting notes from information already present in the clinical context
• Summarizing patient information for review
• Supporting coordination across connected healthcare data sources
• Keeping AI-assisted tasks within the EHR interface rather than a separate workspace

The report is careful to note these are illustrative examples rather than guaranteed capabilities. What any given deployment actually offers will be determined by the health system and the integration it has arranged.

Privacy and data handling

OpenAI's healthcare messaging puts data handling front and center. As relayed by dev.to, the company says data connected to its Health features can be configured so it is not used to train OpenAI's foundation models, and that information synced from connected health sources is deleted from OpenAI's systems within 30 days.

Those commitments matter, but they are not a substitute for evaluating a specific rollout. The report points out that organizations still need to establish exactly which information is connected, who can use the feature, how it behaves inside their existing EHR environment, and which local policies govern staff use. Any AI-generated draft or summary also needs human review before it informs care or documentation, with OpenAI positioning the tool as in-workflow assistance rather than a substitute for clinical judgment.

What remains unanswered

The biggest open question is breadth of availability. OpenAI has confirmed the direction, but per dev.to its initial communications do not identify compatible EHR vendors, regional coverage, user permissions, or commercial terms. Access will depend on enterprise partnerships and deployment-specific arrangements.

The report suggests providers, particularly smaller practices, treat this as something to scope narrowly: start with one well-defined use case such as summarization or documentation support, and check whether the integration actually reduces administrative friction before rethinking wider processes.

Why it matters

Healthcare has been one of the slowest and highest-stakes frontiers for AI adoption, largely because the work is deeply context-dependent and the data is tightly regulated. A generic chatbot, however capable, adds limited value if a clinician has to move information in and out of it manually. Connecting ChatGPT to the EHR changes that equation: summaries and drafts only become useful when they draw on the right patient information, at the right moment, inside the systems care teams already rely on.

The move also signals where the healthcare AI market is heading. Value will come less from standalone chat and more from the quality, safety and availability of connections to real clinical systems. For now the direction is confirmed but the practical details are not, so provider interest should be paired with due diligence on the specific deployment, its data-handling settings and the review steps that surround any AI-assisted output.