Patient intake forms and AI data leakage: what a new patient form gives away in ChatGPT or Claude.

A new patient form is the densest document a practice collects. In one page it holds the patient’s identity, Medicare details, contact numbers, medical history, medications, alcohol and drug use, who to call in an emergency and what they agree to. Reception teams paste completed forms into ChatGPT or Claude to summarise them for the doctor or to fill the practice software. This page marks every item on a routine registration form that the Privacy Act and the NSW health records law bear on, including the patient’s choice about AI, and shows how the same form is read on Ermos inside the practice.

The new patient intake form

Completed by a patient on or before their first visit, then entered into the practice software. Staff often read it before the first consultation to prepare the doctor.

New patient registration
Kestrel Lane Medical Practice · Completed 5 October 2026

Name and date of birth
Address and contact
Medicare
Emergency contact
Medical history
Alcohol, drugs, safety
Occupation
Previous doctor
Consents
Signed

An invented registration form. The Medicare number, phone numbers and email address are placeholders. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    Full name and date of birth

    Two identifiers that tie the form to one person. Personal information collected by a health service provider, and health information once it sits in the clinical record. Privacy Act guide.

  2. 2
    Address, mobile and email

    The details used to reach or find the patient. Personal information, and the material scams and impersonation are built from. Privacy Act guide.

  3. 3
    Medicare number and expiry

    An identifier assigned by an agency, which APP 9.2 limits an organisation from using or disclosing except in listed cases. With a date of birth it is enough to impersonate a patient. Privacy Act guide.

  4. 4
    Emergency contact and who may be told what

    A named partner, a number and an instruction that the parents are told nothing. Third-party personal information, and the patient’s expressed wishes about how information is shared. Patient data guide.

  5. 5
    Mental health history and medication

    A diagnosis-level history, a medication and a prescriber’s location. Health information is sensitive information under the Privacy Act. Patient data guide.

  6. 6
    Alcohol, drug use and self-harm thoughts

    The most private answers on the form, including a safety disclosure the doctor will act on. Sensitive health information, held in confidence under Good medical practice 4.4. Patient data guide.

  7. 7
    Employer and a fear of being found out

    A named employer and the patient’s worry that the manager will learn of the appointment. A confidence that is only safe while it stays in the practice. Privacy Act guide.

  8. 8
    Previous doctor and a records instruction

    A named doctor and a limit on what records to request. Part of the patient’s expressed wishes about their care.

  9. 9
    Consents, including a refusal of AI scribing

    What the patient agrees to, and that they declined AI scribing. The RACGP asks practices to get consent at the start of each consultation using a scribe, and the patient has said no. Patient data guide.

  10. 10
    The signature and date

    Shows the form was completed and when. Evidence of the consents above.

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.

Common questions

Do ChatGPT and Claude use the information in a new patient intake form?
Yes, a pasted intake form 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. One form carries a Medicare number and expiry, a mental health history, answers about alcohol, cannabis and self-harm thoughts, and an instruction that the patient's parents are told nothing.
Is it safe to paste a new patient intake form into ChatGPT or Claude if we remove the name?
No, because the form holds many other identifiers. 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. The street address, mobile number, email address, date of birth and a named employer in North Sydney each point to one person. The diagnoses and medication are health information, which is sensitive 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.
Can I upload a new patient intake form to ChatGPT or Claude instead of pasting it?
An uploaded file leaves the practice as a whole file, which can carry other sheets, hidden columns, comments, tracked changes and file properties as well as the part you meant to share. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Everything on this form travels together, including the refusal of AI scribing, the partner's phone number and the instruction that the parents are told nothing.
How can a medical practice use AI for a new patient intake form 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 summary for the doctor is drafted from the completed form and the practice's own records. A clinician checks it against the form before it goes into the patient record.
Is a Medicare number treated differently from other details on the form?
It is personal information and, because it is an identifier assigned by Services Australia, APP 9.2 restricts when an organisation may use or disclose it. Alongside a date of birth and address it also supports impersonation, which is why it is risky to paste a whole intake form into a tool the practice does not control.
The patient refused AI scribing. Does that cover pasting the form into a chatbot?
The refusal is about the scribe, but the principle is wider. Ahpra says practitioners should inform patients about their use of AI and consider any concerns raised. A patient who declined an AI scribe would be unlikely to expect their registration form to be pasted into a chat tool, and APP 6 turns on what the patient would reasonably expect.
Does an intake form count as health information if it is only registration details?
Yes. Section 6FA(b) of the Privacy Act treats other personal information collected to provide, or in providing, a health service as health information. A name, an address and a date of birth collected on a patient registration form are therefore health information once the practice holds them as a health service provider.

See Dominion answering from your own documents.

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

Book a demo