Americans Are Using ChatGPT as a Doctor. That’s a Problem.

According to Fortune, a new report from OpenAI finds three in five Americans have recently used AI tools for health or healthcare queries, with ChatGPT itself tracking between 1.5 and 2 million health insurance-related messages each week. This surge comes just one week after the expiry of the Affordable Care Act’s enhanced subsidies, a lifeline that helped about 24 million people afford coverage through 2025. The report notes the U.S. healthcare system is a “long-standing and worsening pain point,” driving users to ask ChatGPT for everything from diagnoses to comparing coverage plans and disputing bills. This trend is especially pronounced in “hospital deserts,” where nearly 600,000 healthcare queries to ChatGPT originate weekly from underserved rural areas. Meanwhile, a December Gallup poll found a record low 16% of Americans are satisfied with the healthcare system overall.

OpenAI’s Awkward Positioning Game

Here’s the thing: OpenAI is in a weird spot. They published this report highlighting how indispensable ChatGPT has become for navigating healthcare hell, basically positioning it as an “ally.” But they’re doing this after they had to change their usage policy in November to explicitly walk back the AI’s ability to give medical advice, following lawsuits alleging ChatGPT contributed to suicides. So on one hand, they’re showcasing this massive, organic user trend that proves a desperate need. On the other, they have to legally cover their assets by saying it’s “not a substitute for professional advice.” It’s a classic tech industry dance—celebrate the disruptive use case, but disclaim all responsibility when it goes wrong. The timing, right as a real-world insurance crisis hits, feels almost too perfect for them to study user behavior.

The Real Driver? Systemic Failure

Let’s be clear. People aren’t turning to a chatbot for a cancer diagnosis because it’s cool. They’re doing it because the system has failed them. The report hints at this, but the data is brutal. A KFF survey found nearly 60% of insured Americans had problems using their insurance. The CDC says 27 million are uninsured. Administrative bloat, as noted by researchers, makes everything slower and more expensive. When you’re facing a confusing EOB form or a $5,000 bill for an MRI, and you can’t get a straight answer from your insurer, of course you’ll paste it into ChatGPT. It’s free, instant, and speaks plain English. The alternative is hours on hold or giving up entirely.

Why This Is So Dangerous

But this is where the optimism crashes into reality. A University of Pennsylvania survey found over 60% of Americans think AI health info is at least somewhat reliable. That’s terrifying. Because studies, like one highlighted by UCSF, show these models can be inaccurate and biased, changing recommendations based on a patient’s race or income. ChatGPT is a brilliant pattern-matching engine, not a medical database. It can hallucinate symptoms or treatments with absolute confidence. Using it to decode insurance jargon? Maybe. Using it to self-diagnose chest pain? You’re gambling with your life. The lawsuits against OpenAI show the stakes aren’t theoretical.

A Band-Aid on a Bullet Wound

So what’s the endgame? AI might help some people ask better questions or understand basic terms. But it’s a symptom-treating tool for a disease that requires major surgery. The report mentions people in low-coverage areas could benefit most from real-time info. True. But information isn’t care. It doesn’t replace a doctor, a therapist, or an affordable insurance plan. We’re watching a society, out of pure desperation, adopt a deeply flawed tech solution because the real one is broken. OpenAI gets its viral use case and tons of data. Users get a potentially dangerous crutch. And the underlying crisis—spiking costs, deadlocked senators, and a profoundly unhappy public—just keeps getting worse. That’s not an “ally.” That’s a stopgap on the road to something much scarier.