MW Sorry, 'Dr. Google.' ChatGPT and Claude are now getting into the business of providing health advice.
By Missy Krasner
OpenAI and Anthropic are moving into the complicated world of health information. Will ChatGPT help or hurt patient health?
OpenAI and Anthropic announced earlier this month that consumers can sync health data from their doctor's offices, fitness apps and smartphones via ChatGPT and Claude, respectively, for the first time.
This means that consumers can then use that data to summarize their medical history, explain test results and identify patterns in their own health and wellness practices.
These services are designed to help people take control of their health and navigate care during a time of unprecedented clinician shortages, long wait times to see a doctor and the costs of health-insurance premiums rising to record levels.
Everyone is already talking to ChatGPT about their health - it gets roughly 230 million health and wellness queries every week - just like they have been for years with "Dr. Google" and WebMD. But the launch of dedicated health AI services will make it more personalized and contextual.
This isn't a new idea. I started working on this problem more than 20 years ago at Google $(GOOG)$ and more recently at Amazon (AMZN). Back then, we were trying to solve the problem of personal health information scattered across disparate portals, apps, wearables and doctors' electronic health record systems (EHRs). I was on the team that launched Google Health in 2008. We built Gmail for your medical records - essentially a consumer-owned and mediated "master medical record" that connected various patient portals from your doctor's office, local hospitals, health insurers, retail pharmacies like CVS Health $(CVS)$ and labs like Labcorp $(LH)$.
It didn't work, but we still dreamed of a smarter health search engine. Apple also tried to tackle this problem in 2018 with the addition of health records in the Health app. Microsoft $(MSFT)$ ended its work on a health record in 2019. I worked on the same issue at Amazon as part of the Alexa Health team. We made Alexa HIPAA compliant and conversationally competent to perform simple health tasks, such as, "Alexa, refill my prescription for Ambien at Walgreens off La Paz Road in Laguna Niguel, Calif."
Why did all of these attempts fail? Surprisingly, despite the headlines ("Google, our next Big Brother"), it was not the privacy concerns. I have learned over the years that privacy is often exchanged for utility. People will readily share their data - even when it comes to their health - for things they value, like better service, cheaper prices, and expedient convenience.
To be clear, OpenAI and Anthropic are trying to combat any privacy fears. Both said they are keeping consumers' health information stored separately from other mainstream chats, and they are offering enhanced privacy to protect sensitive data. They also each assert that conversations in health chats will not be used to train the larger foundation models.
The tech industry's past failures in consumer health were not about privacy. They were because of the poor user experience. It really boiled down to three main challenges:
-- Identity and user authentication: This ensures that you are who you say you are. It rationalizes identities so that two people with the same name don't mix up their health data. Over the years, this has gotten a lot easier, but the U.S. healthcare system still does not have a national identifier - like a medical Social Security number - to make it seamless for companies to authenticate your identity.
-- Linking accounts: In order to get data from your CVS or Elevance ELV patient portal, you have to link your accounts. That means you have to first register for an online account. Many people never activate their online portals. If you did register, you have to remember your password and have to sign in to share your account. This process is far easier than it was 20 years ago, as we have modern identity standards like OAuth (Open Authorization), which allows people to grant third-party websites and apps access to their information without giving them their passwords. But it's still time-consuming.
-- Making your data actionable: Perhaps the biggest issue that remains is how to make your medical-record data readable and actionable. That's less of an OpenAI or Anthropic issue and more of a healthcare-system problem. When you finally locate your medical-record data and link it over, it comes across as gobbledygook unless it's cleaned, standardized and mapped correctly. OpenAI is using Bwell to bring in medical records, standardize them and map them for AI to use them. Anthropic is using HealthEx to do the same. But even if you do this well, most of your medical data is still unintelligible or in "doctor speak." That's where OpenAI and Anthropic can help. They can do all that work behind the scenes to make your data easy to understand and useful for everyday care planning.
Bottom line: OpenAI and Anthropic are offering these services to support consumers. It's not to replace physicians. We had the same idea long ago. But if we can leverage the power of AI to do today what we could not do before, the impact could actually help consumers better understand their health.
Think about it. You can share your height, weight, lab work and list of medications to get a response. The chatbot remembers you, creating a more seamless experience than starting another google search from scratch.
There will always be naysayers and laggard adopters, especially those who don't want to give personal data to commercial large language model (LLM) companies. Others will be overly concerned about LLM hallucinations around medical advice. But this is no different than how you search for health answers today. You type in a query, get an answer and then do some more research to verify. You may even double-check it with your doctor.
If these past friction points are addressed, an entire ecosystem will spring up from these new services. Much like Apple's App Store, I expect OpenAI and Anthropic to host third-party apps that will help patients with diabetes, cancer, hypertension, menopause, cardiovascular disease, fitness and weight loss.
Mark my words, the way we search for health information will change forever.
Missy Krasner is a seasoned healthcare executive with 30+ years in the healthcare and digital health industry. She has worked at Amazon, Google and Box, in addition to startups, foundations, the U.S. government, and venture capital. She currently serves as an advisor, board member and investor to several healthcare AI and digital health companies. Most recently, she helped co-found Penguin Ai, a healthcare AI company focused on back-office administration. Missy has a M.A. from Stanford University and a B.A. from UCLA. She lives in Southern California with her husband and is a caregiver to her mother who has multiple sclerosis.
-Missy Krasner
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
January 21, 2026 08:30 ET (13:30 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments