GP CPD Reflection Records: Use case and AI Confidentiality

A doctor has to show continuing professional development that includes reviewing their own performance and measuring outcomes, and the reflection that goes with it is written about real cases. A GP with fifteen minutes between patients pastes the near miss, the audit list and the notes about a complaint into ChatGPT or Claude and asks for a tidy CPD reflection record. The record is meant to describe learning, but it is built from patients, colleagues and a regulator. This page marks every item in an invented CPD reflection record that the Privacy Act, Good medical practice and the NSW health records law bear on, and shows how the same record is drafted on Ermos inside the practice.

The CPD reflection record

Written by the doctor after a performance review activity, an audit or a feedback conversation, and logged with the doctor’s CPD home with the rest of the year’s CPD. Doctors write most of it late in the day, from memory and from the notes.

Wombeyan Street Family Practice
8 Wombeyan Street, Eastwood NSW 2122 · (02) 5550 0161

CPD activity record and reflection · 2026 CPD year · entries made Monday 16 November 2026

Doctor
Professional development plan, written 12 January 2026
Entry 1: reviewing performance, case review (3 hours)
Reflection
Entry 2: measuring outcomes, audit (6 hours)
Changes made
Entry 3: reviewing performance, complaint (1.5 hours)
Entry 4: reviewing performance, colleague feedback (2 hours)
Evidence attached
Hours in this record

An invented CPD reflection record. The practice, doctors, staff and patients are fictional and the registration number is a placeholder. Hover or tap any highlighted item for the analysis; the full list is below.

CPD reflection record drafter

An Agent+ that helps a doctor turn rough notes into a reflection for the year's CPD. It works only from what the doctor supplies, sets the content under your CPD headings, and flags where the learning, the change made or the evidence is missing.

What it does

  1. Reads the doctor's notes and the activity details
  2. Drafts the reflection under the headings the record uses
  3. Flags missing learning points, changes made and evidence

What stays with your people. The doctor owns the reflection, edits it in their own words and decides what is logged. What is said about patients and colleagues stays the doctor's call.

An Agent+ is built once and anyone in the practice can run it. It works from your own documents and runs on Ermos Dominion, the AI unit in your office, or on Ermos Edge, your own instance in an Australian data centre.

What leaves the practice

  1. 1
    The doctor, her registration number and her CPD home

    A register name, a masked registration number and the body that holds her CPD log. The Medical Board’s CPD page says a doctor joins a CPD home and logs activities with it. Harmless alone; it ties everything below to one doctor.

  2. 2
    The professional development plan

    The Board’s page says a doctor writes a plan before starting CPD each year, and Good medical practice 9.2.3 asks doctors to plan and regularly review their CPD. This one records where she thinks she is weak, including a habit and a complaint-related gap. It is her own self-assessment, and she would not want it passed around.

  3. 3
    A named patient, her medicine and the error

    A patient, her age, an anticoagulant, a prescribing error and the hospital that discharged her. Health information is sensitive information under s 6(1) of the Privacy Act, and the Act covers information collected in providing a health service (s 6FA). Good medical practice 4.4.1 asks doctors to treat information about patients as confidential. In NSW, HPP 10 limits using health information for a secondary purpose to consent, a directly related purpose the patient would reasonably expect, and listed exceptions. A reflection rarely needs her name. Patient data guide.

  4. 4
    The open disclosure and the family’s reaction

    A named patient’s son, his anger and a threat to complain, set beside the doctor’s own distress. Good medical practice 8.2.2 asks doctors to be aware of the principles of open disclosure. The son’s reaction is personal information about a second person, and a reflection is a doctor’s own learning, which a draft written by a tool is not. Patient data guide.

  5. 5
    An audit with named patients and their results

    The RACGP describes measuring outcomes as activities that use your own work data. The data here is a patient list with laboratory values. The totals could go into a reflection without the names. The three names and their INR results are health information (s 6FA). Patient data guide.

  6. 6
    The changes made and the repeat date

    A process change and a month. It shows what the practice decided and nothing about any patient.

  7. 7
    A regulator complaint and the doctor’s admission

    A complaint to the HCCC, a deadline, an insurer notification and a candid statement about what she would do differently. Good medical practice 4.12.7 asks doctors to reflect on a complaint and learn from it. The minutes say the insurer has been notified, so check with it before the reflection goes anywhere else. The patient is not named here, and the date and the diagnosis are enough to identify her to the practice. Patient data guide.

  8. 8
    Named colleagues’ feedback

    What a nurse and a registrar said about her. Personal information about two staff members, given to her in confidence in a feedback conversation. It is not patient information, but it is as unwelcome in a vendor’s system.

  9. 9
    Attached meeting minutes and audit spreadsheet

    Evidence that carries more than the reflection does. The minutes name a patient, record a doctor’s health and describe a registrar’s performance. A CPD log needs the activity and has no use for those people.

  10. 10
    The hours and the activity types

    Hours by type, against the Board’s page, which sets 50 hours each calendar year, with 25 hours of reviewing performance and measuring outcomes and five hours minimum of each type. Numbers and labels with nothing identifying.

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 Use cases in general practitioner (GP) practices.

Common questions

Do ChatGPT and Claude use the information in a CPD reflection record?
Yes, a pasted CPD reflection record 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 record carries a named 84-year-old patient on warfarin, an audit naming three patients with their INR results, an HCCC complaint with an insurer notification and feedback from a nurse and a registrar.
Can we use ChatGPT or Claude for a CPD reflection record if we remove the names?
Not safely, because the record stays identifiable after the names go. 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 84-year-old on warfarin, a pharmacist's call, a hospital discharge summary and an angry son are recognisable to the practice, and a complaint about a delayed skin cancer diagnosis with a response date identifies the patient to anyone who knows the case. 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 ChatGPT or Claude be trusted to draft a CPD reflection record?
Not without the doctor checking every line. Language models can produce fluent text with details that are not in the source, wrong figures or dates, and references that look real but are not. In this record the costliest errors would be the audit counts of 46 patients and 31 in range, an INR result against the wrong patient, a moved hospital or date, or hours that do not add up to the 12.5 logged. A reflection is also the doctor's own learning, so a drafted one still has to be rewritten by her.
How can a GP use AI for a CPD reflection record 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 draft is produced from the doctor's own notes, the audit and the practice's template. The doctor rewrites the reflection in her own words and checks every figure before logging it with her CPD home.
How many CPD hours does a doctor need, and what must they include?
The Medical Board's CPD page says doctors need 50 hours of CPD each calendar year. Of those, 25 hours must be reviewing performance and measuring outcomes, with a minimum of five hours of each type, 12.5 hours are educational activities and 12.5 hours can come from any type. A doctor writes a professional development plan before starting CPD each year and, unless exempt, joins a CPD home.
Can a doctor put a real patient into a CPD reflection?
Good medical practice 4.4.1 asks doctors to treat information about patients as confidential. In NSW, HPP 10 limits using health information for a secondary purpose to consent, a directly related purpose the patient would reasonably expect, and listed exceptions, including a training one with conditions about identifying the patient. A reflection rarely needs a name, so de-identifying it avoids the question.
What does Ahpra say about doctors using AI to write up their work?
Ahpra's guidance says the practitioner remains responsible and must apply human judgment to any AI output. It tells practitioners to understand how data is used to retrain the AI, where it is located and how it is stored, and warns that public generative AI tools can make personal information public. The OAIC's guidance says privacy obligations apply to any personal information put into an AI system, and to output that contains personal information.

See Dominion answering from your own documents.

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

Book a demo