In short: you can use ChatGPT, Claude and other AI assistants to understand your health information, but what you paste or connect leaves your device and is processed by the company that runs the assistant, under its own terms. Share only what the question needs, check whether your chats can be used to improve the models, and prefer a connector you can limit and switch off over pasting whole documents.
At a glance
| What you paste | Becomes part of the conversation, is processed by the assistant's company under its privacy policy and usually stays in your chat history |
|---|---|
| Training | Whether conversations may be used to improve the models depends on the company and on your settings |
| Pasting vs a connector | Pasting sends the whole document; a connector can be limited to the parts you choose, shows you what was read and can be switched off |
| Good for | Explaining terms, preparing questions for your doctor, organising your own notes |
| Not for | Diagnosis or treatment decisions: an assistant can be wrong, and it is not a doctor |
Before you share anything: four settings to check
- Training. In the assistant's privacy or data settings, see whether your conversations may be used to improve the models, and turn it off if you prefer. The names of these settings change, so look under the privacy or data controls section.
- Temporary chats. Some assistants offer chats that are not kept in your history. Use one for a one-off health question.
- Memory. If the assistant remembers details between chats, check what it remembers and delete what you do not want kept.
- Connected apps. See which apps and connectors have access, and remove the ones you no longer use.
What not to paste
- Identifiers the question does not need: full name, date of birth, national ID or insurance number, address, patient numbers. For most questions "a woman in her forties" is enough.
- A whole document when one value is the question. Copy just the line, or crop the photo.
- Other people's health information (a parent's, a child's, a partner's) without thinking about whether they would agree.
- Passwords and access codes, including the passcode of a link to your record.
Pasting a document or connecting your record
When you paste a lab report, the whole report goes into the conversation and stays in your chat history. A connector works differently: the assistant asks for data through it, you choose in advance which parts it may read, and you can see the reads and stop them.
The Anpheros Health connector for Claude works this way: read-only by default, only the parts you tick, every read logged in the app, and one tap to stop. What Claude reads still goes to Anthropic, the company that runs Claude, under its terms, which is why the settings above matter either way. The details are in Claude and your health data: privacy, and the setup takes about two minutes: connect Claude to your medical records.
Use the answer well
- Ask it to explain, not to decide: "what does this result mean in general, and what should I ask my doctor?"
- Check important answers against the original document, and with your doctor or pharmacist.
- Never start, stop or change a medicine because of a chat answer.
- If something feels urgent, call your doctor or your local emergency number; do not wait for a chat.
Rights you keep
Under the GDPR (in the EU and the UK) you can ask a company what personal data it holds about you and ask it to erase it; its privacy policy says how. Delete the chats you no longer need.
Frequently asked questions
Is it safe to give ChatGPT my medical records?
It can be reasonable for a specific question if you share only what is needed and check the data and training settings first. What you share is processed by OpenAI, the company that runs ChatGPT, under its terms. Remove identifiers you do not need and avoid pasting whole records.
Do ChatGPT or Claude use my health data to train their models?
It depends on the company and on your settings. Check the privacy or data settings of the assistant before you share health details, and turn off model training there if you prefer.
Can an AI assistant diagnose me?
No. It can explain terms and help you prepare questions, but it can be wrong and it is not a doctor. Diagnosis and treatment decisions are for your doctor.
What is safer: pasting a lab report or using a connector?
A connector you can limit, see and switch off gives you more control than pasting a whole document, which stays in your chat history. Either way, what the assistant reads is processed by the company that runs it.
Can I delete what I shared?
You can delete chats in the assistant, and under the GDPR you can ask the company to erase your personal data. How long deleted chats are kept is set by each company's policy.
Does Anpheros see my chats with Claude?
No. Anpheros does not store your conversations with Claude. It stores what Claude adds to your record after you approve it, and the record of each access.
Try it in Anpheros Daily
Connect your Anpheros record to Claude: you choose what it may read, and you can stop it at any time.
How it works in the app →
Anpheros builds a medical record that the patient controls and can share with any clinic, on an HL7 FHIR R4 platform, based in the EU. These guides help you understand and organise your health; they do not replace your doctor.
About Anpheros →
AnpherosThe more you log, the clearer your record
Your health history in one place, under your control.
Free · Works in your browser · No card
