Patient text messages and AI data leakage: what an SMS thread gives away in ChatGPT or Claude.

Reception teams use text messages to recall patients, confirm appointments and chase results, and patients reply with more than they should. A short thread can hold the patient’s name and number, the doctor, a hint about a test, a fear, a family history and an instruction about who sees the screen. Staff paste threads into ChatGPT or Claude to draft the next message or summarise the contact for the file. This page marks every item in a routine recall thread that the Privacy Act and the NSW health records law bear on, and shows how the same thread is handled on Ermos inside the practice.

The appointment and recall SMS thread

Sent by reception when a doctor asks for a patient to be recalled, usually after a result. Patients reply by text, often from work or from a device other people can see.

SMS thread
· Kestrel Lane Medical Practice

Wed 21 Oct, 9.12am Bec: Hi Peter, it’s Bec from Kestrel Lane Medical. Please call us today on (02) 5550 0142.

Wed 21 Oct, 9.40am Peter:

Wed 21 Oct, 9.52am Bec:

Wed 21 Oct, 10.30am Peter:

Wed 21 Oct, 10.34am Bec:

Wed 21 Oct, 4.05pm Peter:

Wed 21 Oct, 4.11pm

An invented SMS thread. The mobile number and the practice phone number are placeholders. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    The patient and his mobile number

    A name and a number that tie the whole thread to one person. Personal information, and health information because it was collected in providing a health service. Privacy Act guide.

  2. 2
    The doctor and the blood test

    That a named doctor wants to discuss a recent test with a named patient. Information about a health service provided to the patient, which section 6FA of the Privacy Act counts as health information. Patient data guide.

  3. 3
    A worry, a workplace and a shared screen

    Where the patient is, that he is anxious and that his wife can read his messages. The detail that would let someone intercept the next text. Privacy Act guide.

  4. 4
    The practice’s standard line about results

    The practice’s own rule on text messages, with its public number. Generic wording that identifies no patient.

  5. 5
    A guess at the result and a family cancer history

    The patient names the test and the disease he fears, and a parent’s cancer. Health information about him and his father, in his own words. Patient data guide.

  6. 6
    The appointment date, time and doctor

    When and with whom the patient will be seen. Information about a health service to be provided, which section 6FA counts as health information. Patient data guide.

  7. 7
    The staff member and the practice sign-off

    A first name and the practice. It identifies who in the practice handled the patient and where.

  8. 8
    An instruction about how to contact him

    The patient asks the practice to stop texting. His expressed wishes about contact, which the practice has to be able to follow. Privacy Act guide.

What happens in Frontier Models (ChatGPT, Claude)

  • The whole document is transmitted to the provider’s servers. On personal plans the user cannot choose where it is stored or processed, and for the major providers that has generally meant outside Australia, under that country’s legal process. Some business plans offer storage in Australia, but the provider still receives the content.
  • On consumer tiers, prompts can be retained and used to train future models.
  • The provider’s staff and systems can reach the content under the provider’s own policies.
  • Once sent, the practice no longer controls it. Retention and deletion happen on the provider’s terms.

The same document on Ermos

Ermos Dominion

The document is drafted on a unit in your office, from your own templates and the patient record. Every marked item stays on the unit, which keeps working with the internet switched off. Nothing in it is transmitted to an AI provider, so there is no disclosure to analyse: the confidentiality work is the same as for your file server.

Ermos Edge

Your documents, the index and your access boundaries stay on your own instance in an Australian data centre. When a staff member asks for this draft, only their question and the excerpts they are permitted to see go to a named Australian provider, who is contractually barred from keeping the content or training any model on it. Nothing is processed outside Australia.

Frontier Models (ChatGPT, Claude)Ermos EdgeErmos Dominion
Where the document is processedProvider servers, generally offshore on personal plansYour instance in an Australian data centre; answers generated in AustraliaThe unit in your office
Who can read itProvider staff and systems, under policyYour staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothingYour staff only
Is it keptOn the provider’s termsKept on your instance; the provider is barred from retaining itStays on the unit
Can it train a modelOn consumer tiers, yesNo, by contractOnly your own unit learns from it, on your hardware

Part of AI data leakage in healthcare practices. See how this works in practice: Recall letters on Ermos.

Common questions

Do ChatGPT and Claude use the information in an appointment and recall SMS thread?
Yes, a pasted text thread is sent to the AI company's servers and processed there to produce a reply. As at October 2026, whether a conversation on a personal plan (ChatGPT Free, Plus or Pro; Claude Free, Pro or Max) can be used to train the models depends on a setting each user controls. Business plans (ChatGPT Business and Enterprise, Claude Team and Enterprise) and the developer APIs sit under commercial terms, and OpenAI and Anthropic each say they do not train on that content by default (OpenAI, Anthropic). On either product, the thumbs up and thumbs down buttons can send a whole conversation back for training. This thread links a patient's name and mobile number to a doctor, a blood test, a father's bowel cancer, and a request to stop texting because his wife reads the shared iPad.
Is it safe to paste an appointment and recall SMS thread into ChatGPT or Claude if we remove the name?
No, because the thread still carries a mobile number and a story. Removing names rarely makes a document anonymous. Information stays personal information while the person is reasonably identifiable, and the OAIC warns that information can be at risk of re-identification even when it has been de-identified or anonymised. A worksite in Chatswood, a doctor, a father who had bowel cancer and an appointment on Thursday 22 October at 11.40am point to one patient. A message about a named patient's blood test is health information under the Privacy Act. The Office of the Australian Information Commissioner (OAIC) recommends in its guidance on commercially available AI products that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools, because of the significant and complex privacy risks.
What if staff use their own ChatGPT or Claude account for an appointment and recall SMS thread?
The practice cannot see which threads left from a personal account. On a personal plan the user cannot choose where conversations are stored or processed. Some business plans let an organisation choose to store data at rest in Australia (OpenAI), which still leaves the content with the AI company. Both companies can also review content that is flagged for safety or policy reasons. A receptionist who pastes this thread to draft the next text sends a patient's number, his worry about his wife seeing messages and his guess about the result outside the practice's own records, so the practice could not say what was sent.
How can a medical practice use AI for an appointment and recall SMS thread without sending it to ChatGPT or Claude?
Ermos is built so the document stays under your control. On the Dominion AI unit it is drafted inside your office and none of it is sent to an AI provider. On Ermos Edge your documents and index stay on your own instance in an Australian data centre, and only the question and the excerpts the user is permitted to see go to a named Australian provider that is contractually barred from keeping them or training any model on them. The next text is drafted from the patient record and the practice's own message wording. A staff member checks it against the patient's contact instruction before anything goes to the patient.
Is an appointment reminder health information?
Yes. Section 6FA of the Privacy Act counts information or an opinion about a health service provided, or to be provided, to an individual as health information, and also other personal information collected in providing a health service. A reminder that names a patient, a doctor and a time falls within that.
Should staff discuss results by text?
The practice decides its own rule, and many practices say results are not given by text. The thread shows why: the patient can be anywhere, and other people can see the screen. APP 11.1 expects reasonable steps to protect personal information from unauthorised access and disclosure, which includes how the practice contacts patients.
What if the patient asks the practice to stop texting?
The instruction becomes part of how the practice handles that patient's information, and it should be recorded where every staff member sees it. Pasting the thread into an outside AI tool to draft the next message adds a further recipient the patient has not agreed to.

See Dominion answering from your own documents.

A demonstration takes 30 minutes, uses no client data, and comes with no obligation.

Book a demo