Medication incidents and AI data leakage: what an error report gives away in ChatGPT or Claude.

A medication error report has to say exactly what happened, to whom, by whom and why. It is one of the documents a registered nurse most wants help writing, because the words will be read by a manager, a GP, possibly a family and possibly a regulator. It is also a document that names a resident’s drugs and a nurse’s mistake. This page marks every item in a routine medication incident report that the Aged Care Act 2024 and the Privacy Act protect, and shows how the same report is drafted on Ermos inside the service.

The medication incident report

Completed by the registered nurse as soon as a medication error is identified, and reviewed by the clinical manager and the GP.

Medication incident report
Kurrajong Gardens Aged Care · Reported by B. Santos (RN)⁸ · 21 October 2026, 2.15pm

Resident. Mei-Ling Chau, 88, Room 11A. Metastatic breast cancer, palliative. Weight 41kg. Allergy: penicillin.¹

What happened. At the 8am round, Mrs Chau was given 10mg of oxycodone immediate release instead of the charted 5mg. The error was identified at 1.45pm when the afternoon RN reconciled the drug register and found the 10mg tablets had been used.²

Who administered. EN Dana Petrovic, under my supervision. Dana has been with us for three months and completed her medication competency in September. She was managing 22 residents’ morning medications alone because the second RN called in sick.³

Effect on the resident. Mrs Chau was drowsier than usual through the morning and did not eat lunch. Respiratory rate 12 at 10am, 10 at midday. Oxygen saturation 94%. No naloxone given. Her daughter, Wen Chau, was present at 11am and asked why her mother was so sleepy; she was told it may be the illness progressing.⁴

Doctor. Dr Nair contacted at 2pm. Advised to withhold the 2pm dose, monitor hourly, and that no further action was needed given Mrs Chau’s palliative status and the modest overdose. Dr Nair also asked that the 10mg tablets be removed from the imprest.⁵

Family informed. Wen Chau informed at 2.30pm of the error and the doctor’s advice. She was upset that she had been told earlier it was the illness. She asked whether this had happened before; I told her I would find out. Note: there was a similar error with a different resident in July.⁶

Contributing factors. 5mg and 10mg oxycodone stored in adjacent slots in the imprest with similar packaging. Single staff member on the round. The electronic medication chart was offline from 7.40am to 8.20am and paper charts were in use.⁷

An invented report. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the service

The resident, her diagnosis and her weight

A named woman who is dying, her cancer, her weight and an allergy. Health information at its most sensitive. Patient data guide.

Frontier Models (ChatGPT, Claude)	Sensitive health information transmitted, generally offshore on personal plans
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The error

Which drug, which dose, and how long it went unnoticed. The core clinical fact, which the resident and her family are entitled to and a frontier model provider is not. Patient data guide.

Frontier Models (ChatGPT, Claude)	The error transmitted; on consumer tiers it can train future models
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The nurse and her circumstances

A named enrolled nurse, her experience, and the conditions she was working under. Her personal information, in a document that records her mistake. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A named nurse’s mistake transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The clinical effect and what the daughter was told

Observations, and the fact that a family member was given a wrong explanation. The sentence a lawyer would read first. Patient data guide.

Frontier Models (ChatGPT, Claude)	The sentence a lawyer reads first, now on a third party’s servers
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The GP’s advice

A named doctor’s clinical decision. Health information, and the doctor’s professional judgement on a third party’s servers. Patient data guide.

Frontier Models (ChatGPT, Claude)	A doctor’s decision transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The family conversation and a prior error

That the daughter was upset, what she asked, and that a similar error occurred in July. The provider’s own history, disclosed. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A prior error disclosed
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The root cause

Storage, staffing and a system outage. The provider’s admission of systemic causes, written for its own quality process.

Frontier Models (ChatGPT, Claude)	Systemic causes transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The reporting nurse

Ties the report to a named registered nurse with her own professional obligations.

Frontier Models (ChatGPT, Claude)	Transmitted, and the leak is traceable to you
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

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 service 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 resident’s 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 Edge	Ermos Dominion
Where the document is processed	Provider servers, generally offshore on personal plans	Your instance in an Australian data centre; answers generated in Australia	The unit in your office
Who can read it	Provider staff and systems, under policy	Your staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothing	Your staff only
Is it kept	On the provider’s terms	Kept on your instance; the provider is barred from retaining it	Stays on the unit
Can it train a model	On consumer tiers, yes	No, by contract	Only your own unit learns from it, on your hardware

Part of AI data leakage in aged care providers.

Common questions

Do ChatGPT and Claude use the information in a medication incident report?

Yes, a pasted report is sent to the AI company and processed there to produce the 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. It holds the dose error, the enrolled nurse's working conditions and what the daughter was first told.

Is it safe to paste a medication incident report into ChatGPT or Claude?

No, because the report ties a named resident's drug and dose to a named nurse's mistake. 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. The diagnosis, the weight and the oxycodone dose are health information. With the names removed, a single nurse covering 22 residents and a similar error in July would still point to the staff involved.

Does deleting the chat remove a medication incident report from ChatGPT or Claude?

Only partly. Deleting a chat removes it from the user's history, but OpenAI and Anthropic each say they can keep a deleted chat on their systems for up to 30 days, and longer where safety or legal obligations apply (OpenAI, Anthropic). Both products also have memory features that can carry details from one conversation into later ones, and on a personal account each user decides whether they are switched on (ChatGPT, Claude). For this report, a nurse's mistake, a palliative resident's dose and an earlier error could still be held by the AI company after the chat is deleted.

How can an aged care service use AI for a medication incident report 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 registered nurse asks for a draft built from the medication chart, the drug register entries and the progress notes in the service's own incident template. Each statement is tied to its source, and the nurse checks and signs the report.

Is a medication error a reportable incident under the Act?

Not automatically. Section 16 lists the reportable incident types, and a medication error is reportable if it falls within one of them, such as neglect or unreasonable use of force, or if the rules prescribe it. Every medication error belongs in the provider's incident management system regardless, which is where this report was meant to stay.

The nurse is named. Why does that matter more here?

Because the report records her mistake and the pressure she was under. That is the right thing to document inside the service, where it informs supervision and system fixes. On a frontier model provider's servers it is a named professional's error, retained on terms she never saw.

How does Ermos help with incident reporting?

The medication chart, the drug register entries and the progress notes are on the unit in your own building. The report is drafted from them, with each statement tied to its source, and the family and the GP receive the same careful account. Nothing leaves the service to produce it.

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 aged care providers AI for healthcare AI and patient data All professions