Patient questions: privacy and safety

What does the AI see when a patient asks a question?

Only what it needs. To sort a question and to write a general answer, the AI is given the text of the question and nothing else. To draft a reply to a clinical question, it’s given the question and the patient’s first name. It never sees the patient’s record, results, history, medication or contact details.

Does the AI provider keep patient data?

No. The AI provider processes the question to produce its answer or draft, and keeps no patient data afterwards.

Do patients have to agree to this?

Yes. The first time a patient asks a question, the portal explains what happens to it and they must tick I understand and consent to this processing. before it can be sent. The explanation says that their question text, and their first name if the question goes to a clinician, may be sent to the AI provider (based in the US) to sort the question and prepare a reply. It also says that general and admin questions may get an automated answer, and that personal clinical questions are reviewed by a clinician before any reply is sent.

When that explanation is updated, every patient is asked to agree again. It was updated in September 2026, so each patient sees it once more the next time they ask a question.

Can a patient opt out?

Yes. The explanation tells patients they can opt out by contacting the clinic. An owner or practitioner then opens the patient’s record and, on the Profile tab, clears Patient can ask questions in their portal under Patient questions. The patient’s portal stops offering the question box and asks them to contact the clinic directly instead, and Clinicify refuses any new question from them. Their past questions and answers stay visible.

Where are questions and answers stored?

In Clinicify, encrypted. Your team sees questions and drafts in the Q&A Queue, and questions with their sent replies on the patient’s Timeline tab. The patient sees their own questions and replies in their portal. Each time a question is read or answered, it’s recorded in the audit log.

Do emails or texts include the question or the answer?

No. When a clinical question arrives, your team’s email includes only the patient’s initials. When a clinician replies, the patient’s email or text says a reply is waiting and asks them to sign in. The question and answer stay behind portal sign-in.

Can the AI give a patient clinical advice without a clinician?

No. The AI never diagnoses, interprets results or makes a clinical decision. Automatic answers are limited to general information that’s the same for everyone, and anything about the patient themselves goes to a clinician, who decides what’s sent. See What the AI will never answer on its own.

Who signs a clinical reply?

The clinician who sends it. A reply to a clinical question can only be sent by someone with a UK professional registration (NMC, GMC, HCPC, GDC or GPhC) on their profile, and it goes out under their name. Clinicify also keeps the original AI draft for audit, so you can always see what was drafted and what was sent.

What if a patient describes an emergency?

Questions are checked for urgent symptoms before the AI is involved. If one is found, the question goes to a clinician and the patient is told straight away to call 999 or NHS 111 if their symptoms are severe, sudden or getting worse. Even so, the Questions page isn’t an emergency service, and patients shouldn’t wait for a reply there when they need urgent help.

How does this fit with our data processing agreement?

For how Clinicify handles patient data on your behalf, including the AI, see UK GDPR and your data processing agreement and How patient data is protected.