Is it safe to upload medical records to ChatGPT?

What is on a medical document

• your name, date of birth and patient or insurance number;
• diagnoses, treatments, results, and sometimes notes about family history or mental health;
• the doctor's and the clinic's names, which can reveal a specialty (oncology, fertility, psychiatry);
• dates of visits and of admission.

What the AI needs to answer you

To explain "what does a high ferritin mean?" or "what is this term in my letter?", it needs the test names, values, reference ranges and medical terms, plus your age range and sex if they change the reading. It does not need your name, your birth date to the day, your patient number or which clinic you went to.

What can go wrong

Some providers have people read part of the conversations: Google says a subset of Gemini chats are reviewed by humans and kept for up to three years (Google). A medical letter linked to your name in a chat history is the kind of record you cannot recall. The 2025 court order that made OpenAI keep even deleted ChatGPT conversations for several months shows that retention can change after you upload (summary of the order).

A safe way to ask

1. Open the report in Hide personal data before ChatGPT, or paste the lines you have questions about.
2. Tag your name, birth date, patient number and address. Add the clinic and doctor names in "Also replace".
3. Ask with context instead of identity: "woman in her 40s, these are my blood results, what do the flagged values mean?"
4. Paste the answer back if you want it with your name for your own records.

When not to use an AI at all

An AI cannot examine you, and its reading of a result can be wrong. Use it to prepare questions for your doctor, never to decide on a treatment or to rule out a symptom. If something worries you, call your doctor or a medical helpline. For records about someone else, such as a parent or a child, ask their permission or keep them out.

Questions