AI & HEALTH · FOR PATIENTS

AI and Your Health Privacy

People tell chatbots things they would hesitate to say aloud — symptoms, fears, diagnoses, family details. It feels like a private conversation. Legally and technically, it usually is not the protected space a clinic visit is. This page explains, in general terms, what happens to what you type into consumer AI services, why health privacy law often does not cover it, and how to get help from AI while sharing less.

By Dr. Mohamed Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine

✓ Medically reviewed by Mohamed Magdy Ibrahim, MD, PhD — Assistant Professor of Medicine, Transplant Nephrology, University of Maryland School of Medicine · Board certified in Internal Medicine (ABIM) · Last reviewed: November 2026

AI & Health is the MedicalVideos.com section on artificial intelligence in health care. We recommend no commercial product; where we test AI systems, we publish our method and our full results.

In short: In general, the health privacy law most people know — HIPAA in the United States — governs health care providers, insurers, and their business partners. It generally does not cover a consumer chatbot you open on your own phone. What you type may be stored, reviewed for safety or quality, and handled under the service's privacy policy, not medical confidentiality rules. So: strip out identifying details, and never paste records with names or record numbers.

The guide

The picture today

HIPAA's protections attach to covered entities — clinicians, hospitals, health plans — and to the companies that process health data for them. When a hospital offers an AI tool inside its own patient portal, that tool generally operates inside that protected framework. When you open a general consumer chatbot and describe your symptoms, you are a user of a consumer service, governed by that company's terms and privacy policy instead. Other laws may still apply — consumer protection and state privacy laws, and in Europe data protection law — but the shield of medical confidentiality usually does not.

Strengths and limits

What happens to your text depends on the service and your settings. Consumer AI providers commonly store conversations, may use them to improve their systems unless you opt out, and may have samples reviewed by people or automated systems for safety. Policies differ between free and paid tiers, and they change over time. The honest summary is that you cannot know exactly where a given conversation will rest or for how long — which is why the safest strategy is controlling what you put in, rather than trusting what happens after.

In practice

Practical habits that cost nothing: remove names, exact dates, addresses, record and insurance numbers before pasting anything; describe yourself in general terms ('a man in his 60s with kidney disease') rather than as an identifiable person; do not upload documents or result PDFs that carry identifying headers; use any setting the service offers to exclude chats from training; and prefer your health system's own portal tools for questions about your actual record. Periodically delete old conversations you no longer need.

Red flags

Be especially careful in situations where identification carries extra risk: stigmatised conditions, mental health crises, genetic information about you and therefore your relatives, and anything involving another person's health — you cannot consent on their behalf. If a service asks for more identifying information than the task needs, that is a reason to stop. And if a covered provider mishandled your information, you can complain to that organisation and, in the US, to the Department of Health and Human Services' Office for Civil Rights.

Common questions

Is what I type into a chatbot covered by HIPAA?

Generally, no — not when you use a consumer service directly. HIPAA binds health care providers, insurers, and their business partners. A chatbot offered by your hospital inside its own systems is a different situation. This page is general information, not legal advice.

Can the company read my conversations?

Providers state in their privacy policies that conversations may be stored and reviewed, including by human reviewers or automated systems, for safety and quality. Details vary by service and settings. Assume a conversation could be read by someone else, and write accordingly.

How do I ask about a condition without identifying myself?

Ask in general terms and round the details: age decade instead of date of birth, condition names without your name or your clinician's, values without the laboratory's identifying header. You will still get a useful general explanation — all a chatbot can safely give anyway.

Should I paste my full medical record for a better answer?

No. The benefit is small — a chatbot cannot act on your record the way a clinician can — and the privacy cost is real, because records are dense with identifiers about you and often about family members too. Ask record-based questions inside your health system's own tools, or at an appointment.

More in For Patients

Related

Sources

This page is educational content about artificial intelligence and health. It is not medical advice and does not replace your own clinician's judgment. Never use an AI tool in an emergency: if you think you may be having a medical emergency, call your local emergency number (911 in the United States) or go to the nearest emergency department. This page is general information about privacy frameworks, not legal advice.