
AI Virtual Receptionist and Health Data: The Right Questions to Ask
Meet your AI virtual assistant and agent
IGEA AI changes the way your practice communicates with patients: automatic bookings, instant replies on WhatsApp and phone, and a digital experience consistent with your brand.
Quick answer
An AI receptionist for medical practices handles data that may reveal health status: name, phone number, reason for the visit, appointment. Before signing, you need a DPA under Article 28 GDPR, defined retention periods, a list of sub-processors and clear rules on recordings and access.
Why a booking can be health data
The AI does not need to ask "what are your symptoms?". It is enough for the patient to say "I would like a dermatology appointment" for the booking to reveal health information. This is special category data under Article 9 GDPR, even when it arrives inside an ordinary sentence.
This changes the scope of your assessment. You are not buying software that manages names and times. You are entrusting a provider with a flow that may contain health data, audio recordings and information the patient would not share with the first person who happens to answer.
Who is the controller and who is the processor
In most cases:
- the medical practice is the data controller. It decides the purposes, the means, the privacy notice and the legal basis;
- the AI receptionist provider is the data processor, under Article 28 GDPR. It acts on documented instructions and may not use the data for its own purposes;
- cloud, telephony, transcription, voice synthesis and WhatsApp are sub-processors. They must be covered by contracts in the chain.
If the provider cannot say which role it plays, the conversation ends there. This is not a bureaucratic detail: it is the difference between a service and a transfer of responsibility without rules.
The questions to ask
| Question | The answer that should put you at ease | Red flag |
|---|---|---|
| Do you have a DPA under Article 28? | Yes, signed and attached to the contract, with roles and instructions. | "We are compliant" with no document. |
| Which sub-processors do you use, and where is the data stored? | An up-to-date list, primary infrastructure in the EU, any extra-EU transfers covered. | "We have no sub-processors" or no answer about countries. |
| How long do you retain audio, transcripts, appointments and logs? | Separate, configurable periods; audio is not kept forever. | "We keep everything" or "we don't know". |
| Who can listen to a recording? | Access limited by role and logged. | "Anyone on the team can listen". |
| Do you use calls to train models? | A distinction between personal and anonymised data, the legal basis and the option to opt out. | "We use calls to improve the model" with no details. |
| Which technical measures do you apply? | Encryption in transit and at rest, backups, restricted access. | Only "we have antivirus". |
| What happens at the end of the contract? | Data export, deletion within a written deadline, revocation of access. | "The data is kept for statistics". |
| What does the system not do? | No triage, no diagnosis; escalation to a person. | "We do everything". |
The last question is the one nobody puts in questionnaires. Yet it is the most important one in healthcare. An AI that promises to "understand urgency" without a written protocol is not safer: it is less predictable.
Audio recordings: the choice that matters
Recording is convenient: it helps with quality, disputes and training. But it is also the most sensitive data in the flow.
- Don't record if you don't need to. If a transcript and structured fields are enough, ask for audio to be switched off.
- Set a short retention period. Audio should not stay for years. The period must be written down and configured.
- Know who listens. Saying "it is encrypted" is not enough. You need to know who has access, under which role and with what audit trail.
What stays with the practice
A DPA does not transfer everything to the provider. The practice remains the controller and has to do its part.
- Patient privacy notice. The provider's notice does not replace the practice's own.
- Legal basis. Performance of the contract, purpose of care, consent for recordings or for marketing: it must be identified and documented.
- Minimisation. If the reason for the visit is not needed to book, don't ask for it.
- Access and urgent-case protocol. Who sees the dashboard? Who can export? What happens when the AI flags an urgent case?
What IGEA does, in practice
- The DPA is attached to the contract and sets out roles, instructions and sub-processors.
- The primary infrastructure is in the EU; any extra-EU transfers are covered by an adequacy decision, standard contractual clauses or binding corporate rules.
- Cloud, telephony, transcription, voice synthesis and language model providers are bound by Article 28-compliant contracts. The up-to-date list of sub-processors is available on request at privacy@igeacall.com.
- Transfers are encrypted, data at rest is encrypted, backups are encrypted and access is limited to authorised staff.
- Retention is defined for audio, transcripts, appointments and logs. Audio recordings are optional and configurable.
- IGEA does not sell data, does not use it for third-party marketing and does not use it for generalised training. Any service improvement is carried out on anonymised data.
- IGEA does not perform triage, does not make diagnoses and does not replace healthcare staff.
AI Act: the patient must know
From 2 August 2026, EU Regulation 2024/1689 sets transparency obligations for systems that interact directly with people. In practice: users must know they are talking to an AI, unless this is obvious.
For a medical practice, a short notice at the start of the call and in the WhatsApp message is enough: "You are speaking with the virtual assistant of [practice name]. If you prefer, you can ask to speak to a member of staff." It should be coordinated with your privacy adviser.
This article does not replace legal advice.
Checklist before signing
- DPA signed and attached to the contract
- Roles in writing: who is the controller, who is the processor
- Up-to-date list of sub-processors
- Defined retention for audio, transcripts, appointments and logs
- Optional, configurable audio recordings
- Restricted, logged access
- Export and deletion at the end of the contract
- Patient privacy notice and "you are talking to an AI" notice
- Protocol for urgent cases and human escalation
Frequently asked questions
Can an AI receptionist process health data?
Yes, but as a processor and with a DPA under Article 28. The practice remains the controller. Minimisation is mandatory: not all the data the AI could collect is actually needed.
Do patients need to consent to their calls being recorded?
It depends on the legal basis and the context. Recording is not mandatory: if you don't need it, switch it off. If you do use it, there must be a clear notice and a defined retention period. The assessment should be made with your privacy adviser.
Can the provider use calls to train the AI?
Only if the contract and the legal basis allow it. Ask for the distinction between personal and anonymised data and check whether you can refuse the use of data for improvement.
Where is the data stored?
IGEA's primary infrastructure is in the EU. If a sub-processor operates outside the EEA, the transfer must be covered by an adequacy decision, standard contractual clauses or binding corporate rules. Ask for the up-to-date list and the countries involved.
What happens to the data at the end of the contract?
It must be exportable and deleted within a written deadline. Always ask for written confirmation.
Can the AI perform triage or give clinical advice?
No. IGEA is not a medical device and does not make diagnoses. It can handle operational information, bookings and administrative requests; for urgent cases and clinical questions it follows the practice's protocol and hands over to a person.
How do I tell patients they are talking to an AI?
With a short notice at the start of the call and in WhatsApp, plus a link to the practice's privacy notice. The notice must be understandable, not a thirty-second legal formula.
If you want to see first how to assess the operational flow, read How to Test an AI Virtual Receptionist. To discuss your own case, contact us.
Meet your AI virtual assistant and agent
IGEA AI changes the way your practice communicates with patients: automatic bookings, instant replies on WhatsApp and phone, and a digital experience consistent with your brand.
Related articles

Revenue and Missed Calls
Virtual Switchboard and AI Receptionist: How a Medical Centre Recovers Lost Revenue
A virtual switchboard with an AI receptionist answers 24/7, books autonomously and reduces missed calls. Here is how a medical centre recovers the revenue that currently slips away over the phone.

Health Technology
Why Choose an AI Receptionist Specialised in Medical Centres
A generic AI receptionist does not necessarily understand a medical practice's workflows. Discover why a specialised solution can handle bookings, patient requests and healthcare communications better.

Operational Efficiency
How to Reduce No-Shows in Medical Practices with an AI Receptionist
Reminders, confirmations and smart call handling help medical practices and clinics reduce gaps in the calendar.