Millions of Americans are turning to artificial intelligence for health advice, and some are now uploading their complete medical records to get personalized AI-powered insights. A March poll from the Kaiser Family Foundation (KFF) found that 1 in 3 American adults have consulted AI about their health, with more than 300 million people worldwide using ChatGPT each week for health questions. OpenAI recently launched a feature allowing US users to upload medical records directly into ChatGPT, enabling the AI to review doctor's notes, lab results, and data from wearable devices to provide more tailored health guidance.

Why Are People Uploading Their Medical Data to AI?

The appeal is clear: AI can synthesize vast amounts of health information quickly. According to the KFF poll, 41% of people who asked AI health questions uploaded their personal medical information to get better answers. The reasons vary. Some want to be better informed before seeing their doctor, while others use AI because they lack access to affordable medical care. As of 2023, an estimated 100 million Americans are considered "medically disenfranchised," meaning they can't reach a primary care doctor because too few practice in their community.

AI companies have made uploading easier by integrating with popular health platforms. OpenAI, Microsoft (which makes Copilot Health), Anthropic (which makes Claude for Healthcare), and Perplexity all now allow users to connect their medical records and Apple Health data directly to their AI tools. The companies promise encrypted conversations, no use of your data for training their models, and the ability to disconnect your records at any time.

What Are the Privacy Risks You Should Consider?

Despite company promises, significant privacy concerns remain. Unlike hospitals and doctors' offices, which must follow strict federal regulations protecting patient health information, AI companies operate in a regulatory gray zone. The United States has no comprehensive federal privacy law governing AI and health data. This creates vulnerability.

A cautionary example: 23andMe, the genetic testing company, experienced a massive data breach in 2023 that exposed personal and genetic data of nearly 7 million customers. When the company filed for bankruptcy in 2025, it sold the genetic data of 15 million customers to a nonprofit created by its founder, with no legal mechanism to prevent that sale. This illustrates how health data can be used in ways you never intended, even after a company promises to protect it.

"The legal and ethical infrastructure to handle all of this information and to handle personal and sensitive information, that just doesn't exist yet," said Dr. Art Caplan, a bioethicist and professor emeritus at the NYU Grossman School of Medicine.

Dr. Art Caplan, Bioethicist and Professor Emeritus, NYU Grossman School of Medicine

How to Decide Whether Uploading Your Medical Data Is Right for You

• Assess Your Risk Tolerance: Understand that while companies promise privacy protections, the risk of a data breach or unexpected use of your information is not zero. Dr. Bob Wachter, chair of the Department of Medicine at the University of California, San Francisco, noted that companies have strong business incentives to protect your data, but "the risk is not zero; it's just very small, and you really need to decide whether it's worth it".
• Verify Before Trusting: Don't assume company privacy policies are foolproof. Dr. Girish Nadkarni, a nephrologist at Mount Sinai who studies AI in healthcare, recommends users "verify, then trust" and seek independent verification that companies don't train on patient data, rather than relying solely on their stated policies.
• Consider Alternative Options: Some electronic medical record providers like Epic, which makes MyChart, are developing their own AI products. If privacy is your primary concern, waiting for these integrated solutions may reduce the need to share data with third-party AI companies.

Can AI Actually Give You Reliable Health Advice?

Even if privacy weren't a concern, the accuracy of AI health guidance remains questionable. While OpenAI claims its latest ChatGPT version has been evaluated by hundreds of clinicians and performs at or better than human clinicians, real-world studies tell a different story. A 2023 study published in JAMA Oncology found that an older version of ChatGPT provided inappropriate recommendations for cancer treatment in over one-third of test cases. Another study published in Nature Medicine in February found that OpenAI's health tool made mistakes and downplayed certain life-threatening circumstances.

The problem is partly how AI works. These programs are built to sound polite and convincing, even when they're wrong. They can "hallucinate" or fabricate information that sounds plausible. They also have documented biases about race and gender that influence their advice. In some cases, AI may tell someone to stay home when they should seek emergency care, a potentially dangerous error.

Unlike a Google search, where you get multiple answers to compare, AI gives you a single response with an air of certainty that makes it hard to detect when the advice is incorrect.

What Does the Future of AI in Healthcare Look Like?

Despite these challenges, AI is becoming increasingly integrated into healthcare systems. Researchers at Georgia Tech note that AI is moving beyond administrative tasks and beginning to support clinical decision-making. AI tools like OpenEvidence now provide healthcare professionals with source-backed answers to medical questions, and AI is being used within electronic health records to generate clinical notes and summarize complex patient charts.

However, experts emphasize that the future should not be about autonomous AI making decisions alone. "The real opportunity lies not in building more autonomous AI, but in designing systems that strengthen human judgment, clinical expertise, and community care," explained Munmun De Choudhury, a professor in the School of Interactive Computing at Georgia Tech. Similarly, Jon Duke, director of the Center for Health Analytics and Informatics at the Georgia Tech Research Institute, noted that "medicine has clearly embraced the human-AI fusion approach. The use of AI is encouraged, but reliance on AI is not".

AI is showing particular promise in mental health, where randomized controlled trials have demonstrated that some people are willing to receive therapy from an AI agent and find it beneficial. Researchers are also exploring how AI can identify emerging mental health concerns by analyzing digital and clinical data, potentially enabling earlier intervention before someone reaches a crisis.

The bottom line: AI can be a useful tool for understanding your health and accessing information when you lack other options, but it should not replace conversations with your doctor. If you do choose to upload your medical data, do so with eyes open to the privacy risks and the limitations of AI accuracy. As Dr. Caplan advised, "I would never turn my data over to a private entity without more assurance of confidentiality or understanding of how they're going to protect against leaks and more clarity about what happens if there is a hack or an accidental leak".