Quality improvement spreadsheets and AI privacy: what a GP clinical audit sheet gives away in ChatGPT or Claude.

Practices run quality improvement work from spreadsheets: an export of every patient with diabetes, their latest results and whether they are due for review. The practice manager or a nurse uploads the sheet to ChatGPT or Claude to find the overdue patients, chart the results or fix a formula. The sheet is a clinical register for a whole cohort, with identifiers. This page marks a routine audit sheet and shows how the same work is done on Ermos inside the practice.

The quality improvement spreadsheet

Exported from the clinical software by the practice manager or a nurse to plan recalls and track quality measures, then shared by email, which is how it reaches an AI tool.

fx
ABCDEFGH
1PatientDOBIndigenous statusHbA1c %eGFRSmokerLast reviewNote
2
3
4
5
6
7
8
9
Diabetes cohort

An invented spreadsheet. All patients are fictional. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    Patient names

    A name on a list titled for diabetes is already a diagnosis. The names were collected in providing a health service, so they are held as health information under section 6FA. Privacy Act guide.

  2. 2
    Dates of birth

    With the names, enough to identify each patient with certainty. Privacy Act guide.

  3. 3
    Aboriginal and Torres Strait Islander status

    Racial or ethnic origin is sensitive information under section 6(1) of the Privacy Act. Privacy Act guide.

  4. 4
    HbA1c and kidney results

    Laboratory values against every named patient in the cohort. Health information, which is sensitive information. Patient data guide.

  5. 5
    Smoking status

    A health behaviour against each name. Patient data guide.

  6. 6
    Last review dates

    When each patient was last seen, and which ones have dropped out of care.

  7. 7
    The notes column

    Free text with the most candid detail: a lost job, an alcohol history, an eating disorder kept to the GP, a carer at home. The column most likely to cause harm. Patient data guide.

  8. 8
    The review formula

    A date test with no patient information in it. The formula can be shared; the sheet it sits in cannot.

  9. 9
    Other tabs and file properties

    The workbook carries a second tab covering every chronic disease patient, plus any hidden columns, comments and properties such as the author. Uploading the file sends all of it.

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: Care plans on Ermos.

Common questions

Do ChatGPT and Claude use the information in a quality improvement spreadsheet?
Yes, an uploaded or pasted sheet 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 tab lines up eight patients with diabetes against their results, smoking status, Indigenous status and notes about money, alcohol and an eating disorder.
Is it safe to upload a quality improvement spreadsheet to ChatGPT or Claude if we delete the name column?
No, because every row keeps a date of birth, results and a note. 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. Indigenous status is sensitive information under the Privacy Act, and a note about a patient's lost job identifies her to anyone who knows her. 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 quality improvement spreadsheet to ChatGPT or Claude instead of pasting it?
An uploaded spreadsheet leaves the practice as a whole file, which can carry other sheets, hidden columns, comments 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. This workbook has an All chronic disease tab beside the diabetes cohort, and both could travel together.
How can a GP use AI for a quality improvement spreadsheet 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 work is done over the practice's own export on the unit, so the overdue list and charts are produced inside the practice. The practice manager or a nurse checks the results before acting on them.
Is a quality improvement export covered by the Privacy Act?
Yes. Every row is a patient's health information, and Indigenous status is sensitive information in its own right. APP 11.1 requires reasonable steps to protect it, which includes deciding where the file is sent.
Would a breach of this sheet be notifiable?
Unauthorised disclosure of personal information likely to result in serious harm is an eligible data breach under section 26WE of the Privacy Act. A cohort list with results and candid notes is the kind of file where serious harm is easy to see.
Can we de-identify the sheet first?
Removing the name column rarely makes the sheet anonymous. Dates of birth, results and notes still point to individuals, especially in a local practice.

See Dominion answering from your own documents.

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

Book a demo