I'm sure other healthcare journalists, like myself, have received pitches from public relations professionals since OpenAI announced a new tool called ChatGPT Health a few days ago. Much of this hype essentially suggested that the generative AI company's announcements fell short of expectations.
For example, one campaign promoted the views of the CEO of an AI-powered health navigation company. The blurb says: Americans don’t need artificial intelligence to help them burn more calories on Peleton (sic). They need AI to help them find care at the lowest price.
The truth is that both statements are correct. People worry about rising medical costs and They also want to stay healthy and understand their health data. This latter reality was undoubtedly the catalyst that led to the birth of ChatGPT Health.
But the question is, why aren't traditional EHR companies or companies that have been innovating with AI in healthcare doing the same themselves?
In fact, I'm not just thinking about companies like Epic Systems, but also companies like Microsoft. Why Microsoft? Because a few years ago, Microsoft demoed a tool that seemed to get to the heart of ChatGPT Health's goal — to provide insights about personal medical records.
It was the fall of 2023, and Microsoft was pitching various AI capabilities to a group of reporters, including me, at an industry event. I volunteered to give a live demonstration of the company's ambient artificial intelligence capabilities, which were brought about by the company's acquisition of Nuance. But the most striking moment of the afternoon came when Linishya Vaz, principal program manager for Microsoft's Health and Life Sciences Group, highlighted a less shiny feature. I'm pasting what I wrote in that story verbatim below:
Imagine you have had some blood tests or radiology images done. The language of radiologists’ reports is filled with scientific jargon—almost intentionally designed to keep patients in the dark, or to guess what the reports are about. I did Google searches to decipher the meaning of imaging and other lab reports, as I'm sure countless others have done over the years.
Microsoft's goal is to make it easier for patients to do this, assuming of course that their imaging provider/health system is a Microsoft customer and has already deployed these capabilities.
In the hypothetical example Vaz shared, a patient with chest pain underwent a chest X-ray. The report reads:
“The left anterior chest wall dual-lead pacemaker is more stable than before
take an exam. Lungs hyperinflated to clear. No pneumothorax or pleurisy
Escape. Pulmonary vessels were normal. The heart is of normal size. bony
The structure is demineralized but still intact.
Impression: Hyperinflation of the lungs, consistent with emphysema. Osteopenia. “
Now, Microsoft's generative AI feature can simplify scientific terms into plain English. And, according to Vaz's demonstration, create a report that patients can view in the patient portal.
* As shown on the previous exam, a device with two leads (a dual-lead pacemaker) is positioned steadily on the left front of your chest.
* Your lungs are overinflated, which is consistent with emphysema
* No air (pneumothorax) or fluid (pleural effusion) around the lungs
* The blood vessels in your lungs look normal.
* Your heart is of normal size
* Your bones lose density (demineralize), but they remain intact. This is called osteopeniaIn summary:
Your lungs are overinflated, consistent with emphysema, and you have osteopenia (loss of bone density).
Patients will also see this very important disclaimer at the bottom of the report: This simplified version was generated by Assistant Al.
The power of this technology in helping patients interpret radiology or lab reports—assuming of course that the generative AI is accurately simplifying—cannot be overestimated. It will completely eliminate the need for patients to conduct time-consuming web searches to find out what is happening to them. And make them feel empowered.
This is what I wrote in October 2023
Now, let’s look at the recent OpenAI announcement, specifically a few sentences from it: There are countless stories of people using ChatGPT to help make sense of it all. In fact, health is one of the most common ways people use ChatGPT today: According to our de-identified analysis of conversations, more than 230 million people around the world ask health and wellness-related questions on ChatGPT every year. Week. (emphasis added)
“…to help make sense of it all.” Thus, ChatGPT Health was born, currently only available to a handful of users.
Now, how will ChatGPT Health make money for OpenAI? I have no idea. Can you trust ChatGPT Health to be HIPAA compliant? That's what they said. But assuming the average ChatGPT user understands the standard, will they be concerned about HIPAA for healthcare? I'd be willing to bet “no.” While this is no excuse to relax privacy standards, the point is simple: People are hungry for a well-designed, consumer-facing tool that allows them to better understand a person's medical records and health trends. People need to centrally store health data, including medical records and wearable devices, and be able to query this collective and disparate data and get easy-to-understand, actionable responses in plain English. OpenAI saw this need and jumped at the opportunity. Microsoft saw this but never expanded on it enough – or so it seems.
When asked if this was a missed opportunity, a spokesperson outside Microsoft forwarded me a blog post written by Microsoft AI CEO Mustafa Suleyman in fall 2025 that included this paragraph about the Washington company's capabilities in consumer-facing health.
Copilot for health solves one of the most common user needs: health-related problems. We've improved the way we respond to trusted sources like Harvard Health to provide users with reliable information. Copilot also helps you find the right doctor quickly and confidently, matching based on specialty, location, language and other preferences. The goal is simple: help you take control of your health by providing you with high-quality information and getting you the right care quickly.
Except Health Co-Pilot isn't really a dedicated health tool at all. It's just Copilot, where you ask health-related questions to a generative AI tool. It does not have the functionality or privacy standards to handle personal medical records or wearable device data.
In other words, this is a huge missed opportunity for Microsoft, as well as traditional healthcare companies that claim to be patient-centric.
Two Epic executives, including the EHR company's chief medical officer, disputed that conclusion. Stay tuned for this story and their thoughts on ChatGPT Health.
Photo: Huang Yuliang, Getty Images
