According to a recent Tebra report, 60% of Americans have used an AI chatbot like ChatGPT, Gemini, or Claude to get health advice. Experts link Americans growing reliance on AI chatbots for health advice to the fact that they offer something traditional healthcare providers cannot — unlimited access at no cost. 

It’s an offer that’s hard to refuse, and chatbot developers know it. For example, ChatGPT recently launched ChatGPT Health which allows consumers to securely connect medical records and wellness apps to better understand their health information. But unfettered access to AI healthcare advice raises many very real safety and privacy concerns that physicians must learn to navigate. 

In its Top 10 Health Technology Hazards for 2026 report, ECRI named the misuse of AI chatbots in healthcare as the number one hazard. The report cited risks associated with  incorrect responses and hallucinations that could jeopardize patient safety and cause harm. The takeaway is clear — using ChatGPT in healthcare requires guardrails and skepticism. This is true for patients and also for physicians.

Let’s take a closer look at how physicians can help keep patients safe as the prevalence of AI chatbots in healthcare grows. 

Patients use ChatGPT in healthcare to quickly find information on medical conditions or treatments, including how to use certain medical devices and what medical supplies to buy. But the answers provided don’t take an individual’s health history into account and aren’t vetted by medical professionals. This means physicians must be ready and willing to steer patients toward safer, more informed care decisions. Here are some strategies that can help transform the patient-provider dynamic in productive ways.

1. Ask patients about their use of AI chatbots for health information. Consider incorporating a question about patients using ChatGPT in healthcare as a standard part of the intake process across specialties. If patients do use it, consider inviting them to share how so you can normalize the behavior and surface any misinformation. If a patient presents options suggested by a chatbot, review the information together and work to  interpret, personalize, and validate what may or may not apply to their specific case.

2. Urge patients to scrutinize responses. Common large language models (e.g., ChatGPT, Claude, Copilot, Gemini, and Grok) are not designed or regulated for healthcare purposes and responses are based on large datasets — not genuine comprehension of medical information. This means patients using ChatGPT in healthcare must exercise caution when reviewing AI-generated information. Physicians should be prepared to explain why AI-generated information may not apply because of the patient’s specific circumstances — including comorbidities, medications, or other factors.

3. Set clear boundaries. Emphasize the limitations of using ChatGPT in healthcare. Let patients know that while chatbots may be helpful for general health education, actual medical decisions should be based on their medical history, exam, and labs.

4. Educate on privacy and security concerns. There is currently no federal regulatory authority overseeing the health information that’s shared with AI chatbots. Additionally, ChatGPT delivers technology services that do not fall under the purview of HIPAA regulations. This means that if a data breach occurs, consumers using ChatGPT in healthcare would not be afforded specific protections or rights under HIPAA. Let patients know that in the absence of greater consumer protection, it’s best for them to query chatbots using general information and not data from their medical records verbatim.

5. Create a patient resource. Physicians can demonstrate their expertise and add value by answering common patient questions about ChatGPT, including clear guidance on how to use it. Here’s a sample guide for patients on using ChatGPT in healthcare.

While AI tools like ChatGPT can be helpful for learning — they cannot replace your healthcare team. Use this guide to stay safe and get the most out of your visits.

What ChatGPT can do

What ChatGPT can’t do

Protect your privacy 

Never enter personal or identifying health information into public AI tools, including:

How to safely use AI tools

Safe uses

Not safe uses

Bring it to your visit

It helps to share what you read and learn with your physician:
“I looked this up using AI and have questions — can you help me understand what applies to me?”
Your care team can explain what fits your situation and what doesn’t.

When to seek care right away

Get medical help promptly for chest pain, trouble breathing, new weakness or numbness, severe or worsening pain, fainting, high fever in infants, or symptoms that are rapidly getting worse.

5 simple rules

Bottom line: AI tools can help you learn about health topics — but only your physician and healthcare team can safely apply that information to you.

As the use of AI in healthcare continues to grow, physicians must become the trusted human layer between AI information and real medical decisions. And while patients using ChatGPT in healthcare poses challenges for physicians, it also presents opportunities for practice growth. This is especially true in cases where chatbots increase anxiety among some patients. By offering same-day sick appointments and telehealth options, practices can retain patients who typically seek care at urgent care or retail clinics. Additionally, practices can support higher reimbursement rates when patients come to appointments with long lists of AI-derived questions that require more complex and lengthy visits.