Discharge summaries and AI patient privacy: what a hospital discharge summary gives away in ChatGPT or Claude.

Discharge summaries arrive in a GP’s inbox every day, often several pages long, with the important actions buried among the hospital’s own notes. GPs and nurses paste them into ChatGPT or Claude to pull out the medication changes and the follow-up the hospital wants. The summary is an admission in one document, written by someone else and held by the practice. This page marks every item a routine summary carries and shows how the same summary is condensed on Ermos inside the practice.

The hospital discharge summary

Sent by the hospital to the patient’s nominated GP when the patient is discharged. The practice files it against the record and actions the follow-up.

Parramatta River Private Hospital
Discharge summary · Discharged 1 November 2026

To: Dr Rafael Szabo, Wombeyan Street Family Practice

Patient:

Admission.

Procedure.

Medication changes.

Other issues.

Social work.

Requests to GP.

Driving.

An invented discharge summary. The hospital, doctors and patient are fictional, the MRN is invented and the Medicare number is a published test value. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    Name, date of birth, hospital number and Medicare number

    Three identifiers, one from the hospital and one from a government agency. APP 9.2 limits when an organisation may use or disclose a government related identifier. Privacy Act guide.

  2. 2
    The admission and diagnosis

    A heart attack against a named patient. Health information is sensitive information under the Privacy Act. Patient data guide.

  3. 3
    The procedure

    What was done in hospital and when. Patient data guide.

  4. 4
    Medication changes

    The list the GP most needs, and the reason the summary is pasted into a tool in the first place. Patient data guide.

  5. 5
    Alcohol use and withdrawal

    A drinking history and a refused referral. Among the most sensitive entries a record can hold. Patient data guide.

  6. 6
    A partner’s report of aggression

    Information from a third party, recorded on the condition that the patient is not told. If it reached the patient by any route it could put the partner at risk. Patient data guide.

  7. 7
    The hospital’s requests to the GP

    Follow-up actions with no identifying detail of their own, though they describe this patient’s care.

  8. 8
    Driving advice

    A restriction tied to the patient’s condition. Patient data guide.

  9. 9
    The hospital doctors

    Names of the treating team, linked here to this admission.

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 general practitioner (GP) practices. See how this works in practice: The GP inbox on Ermos.

Common questions

Do ChatGPT and Claude use the information in a hospital discharge summary?
Yes, a pasted or uploaded summary 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 summary holds a heart attack, a stent, five medicine changes, a drinking history and a partner's report of aggression that the patient has not been told about.
Is it safe to paste a discharge summary into ChatGPT or Claude if we delete the patient's details?
No, because the hospital number, dates and clinical story still point to one man. 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. An admission date, a procedure and a partner's report identify him to anyone at the hospital or at home. 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 discharge summary PDF to ChatGPT or Claude instead of pasting it?
An uploaded file leaves the practice whole, with every page, the hospital's headers and any attached results, as well as the part you meant to summarise. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. The social work note on page two travels with the medicine list on page one.
How can a GP use AI for a hospital discharge summary 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 is condensed from the document as it sits in the practice inbox, with medicine changes listed against the record. The GP checks the summary against the source before accepting it.
The hospital sent the summary to us. Can we do what we like with it?
The practice holds the summary as health information about its patient, so the Privacy Act applies to it as it does to the practice's own notes. Using it to provide care is the purpose it was sent for; sending it to an AI provider is a separate disclosure.
Why is the social work note so sensitive?
It records a partner's report of aggression on the condition that the patient is not told. Any disclosure that could reach the patient puts the partner at risk, so it is the item the practice most needs to keep under its own control.
Can AI safely pull out medication changes from a discharge summary?
It can draft a list, which the GP then checks against the source and the current record. Medication reconciliation remains the GP's responsibility, wherever the draft is made.

See Dominion answering from your own documents.

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

Book a demo
Keep reading
AI data leakage in general practitioner (GP) practices AI for general practitioners (GPs) AI and patient data All professions