Healthcare Clinical Photo Registers: Use case and AI Confidentiality

After a tip-off that doctors were keeping patient photos on personal phones, the practice manager audits every clinical image taken in a month and lists them in a spreadsheet: who was photographed, what the picture shows, who took it, what consent was recorded, where the file sits now and who it was sent to. She then pastes the sheet into ChatGPT or Claude to draft a clinical photography policy. Each row names a patient and a condition. This page marks every item in a routine clinical photo register that the Privacy Act, NSW health records law and Good medical practice bear on, and shows how the same sheet is handled on Ermos inside the practice.

The clinical photo register

Compiled by a practice manager or nurse when auditing the clinical photos taken in the practice, usually before writing or revising a photography policy. The RACGP suggests such a policy cover how photos are produced, consented to, uploaded to the record, removed from personal devices, retained and disposed of.

fx=COUNTIF(E2:E10,"*personal*")
ABCDEFGH
1RefTakenPatientPhotographedTaken by, deviceConsentNow storedSent to
2Sam Kovacs RN, practice tabletWritten form signedPatient record onlyNot sent
3
4Dr Mirchandani, practice tabletRecorded in the notePatient record onlyNot sent
5
6Dr Voss, practice cameraWritten form signedPatient record onlyNot sent
7
8Dr Voss, practice tabletRecorded; asked for no sharingPatient record onlyNot sent
9Dr Voss, practice tabletRecorded in the notePatient record onlyNot sent
10Sam Kovacs RN, practice tabletWritten form signedPatient record onlyNot sent
11AuditCompiled Fri 13 Nov by Helen Tran9 images
Photo auditPolicy draft

An invented register. The patients are fictional and the dates are in 2026; the practice and doctors are the same invented cast as the other healthcare documents. Hover or tap any highlighted item for the analysis; the full list is below.

Clinical photo audit register builder

An Agent+ that builds a clinical photo register from the audit notes and consent entries your practice supplies. It lists each image, where it is held and what consent is recorded, and compares each row against the text of your photography policy.

What it does

  1. Reads the audit notes and recorded consents
  2. Builds the register one image per row
  3. Flags images held outside the patient record

What stays with your people. The practice manager or nurse checks each flagged row and decides what is kept or removed. The agent builds the register; people act on it.

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
    Reference numbers and dates

    The practice’s own sequence numbers and the day each image was taken. Alone they identify no one. Next to the names in the same row they place a particular patient at a particular appointment.

  2. 2
    Patient names and ages

    A name and an age identify a person. Linked to a photographed condition they are health information, which includes other personal information collected in providing a health service (Privacy Act s 6FA(b)) and is sensitive information under s 6(1). A private business that provides a health service and holds health information is not within the small business exemption (s 6D(4)(b)). Patient data guide.

  3. 3
    What each image shows

    A wound, a rash, a mole, a burn, swollen ankles and a face. Each is health information, and some pictures show more than the condition: a tattoo and part of a face in one frame, an abdomen, a child. The RACGP says clinical photos taken for patient management are part of the patient’s health record, even if they only exist electronically. Good medical practice 4.4.1 asks doctors to treat information about patients as confidential. Patient data guide.

  4. 4
    Personal phones

    Identifies which doctor or nurse used a personal phone. The RACGP says a clinician’s phone might not be as secure as the practice’s own devices, is more likely to be lost, stolen or accessed by others, and that a clinician might forget to transfer or delete the photo. HPP 5(1)(c) in NSW asks for security safeguards that are reasonable in the circumstances, and Good medical practice 10.5.2 asks doctors to hold records securely and protect them against unauthorised access. The entry also describes a named employee’s practice. Patient data guide.

  5. 5
    Consent that was verbal and not written down

    Good medical practice 4.5 describes informed consent as a voluntary decision made with knowledge and understanding of the benefits and risks, and 4.4.4 asks doctors to use consent processes for the release and exchange of health information. APP 3.3 asks an organisation for the individual’s consent to collect sensitive information that is reasonably necessary for its functions, unless APP 3.4 applies, which includes a permitted health situation (s 16B(1)). HPP 4 asks the organisation to take reasonable steps to make the person aware of why it collects the information and who it is usually given to. A verbal consent that is not recorded is hard to show later. Patient data guide.

  6. 6
    Camera roll, cloud backup and a messaging group

    Says where the image sits now. The RACGP key points are to store photos in the patient’s health record, always delete them from the device, and avoid third-party storage and automated back-ups. HPP 5(1)(a) and (b) ask that health information is kept no longer than necessary and disposed of securely, and APP 11.2 asks an entity to destroy or de-identify personal information it no longer needs unless a law requires it to be kept. Patient data guide.

  7. 7
    A photo sent to a specialist by WhatsApp

    Sending a mole photo to a dermatologist to help a patient’s care is the kind of disclosure HPP 11(1)(b) and APP 6.2(a) can allow, because the purpose is directly related to the reason the information was collected and the patient would reasonably expect it. The channel is the problem. The RACGP says photos can be sent to the wrong person or intercepted, and there is no guarantee they will be stored securely on the recipient’s device. Good medical practice 4.4.6 asks doctors to ensure that digital communications are consistent with their duty to protect patient confidentiality. Patient data guide.

  8. 8
    A child’s rash in a group of seven

    A six-year-old’s photo, shared with everyone in a staff group, most of whom do not treat her. The mother agreed verbally to a photo for the doctor, and nothing says she agreed to seven people seeing it. Good medical practice 4.4.3 asks doctors to access a medical record only when there is a legitimate need, and HPP 10(1) and 11(1) limit use and disclosure of health information for a purpose other than the one it was collected for. Patient data guide.

  9. 9
    An employer on copy

    A burn photo, with a tattoo and part of a face, sent by email to an insurer’s case manager and copied to the patient’s employer. The patient agreed to a photo for a claim. Whether he agreed to the employer seeing it is not recorded. HPP 11(1)(a) and APP 6.1(a) allow a disclosure for another purpose with the person’s consent, and the practice should check what the patient agreed to before relying on any other basis. This page does not set out the workers compensation rules on what an employer may see. Privacy Act guide.

  10. 10
    The audit counts and the formula

    Counts rows by device, consent, storage and sending. The COUNTIF formula names a column and a word and holds no patient data. The counts show the practice has a gap (three of nine images on personal phones) and a model could mislabel the rows it counts.

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 healthcare practices. See how this works in practice: Clinical documentation on Ermos.

Common questions

Do ChatGPT and Claude use the information in a clinical photo register?
Yes, whatever is pasted or uploaded from a register 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 register links nine patients by name and age to what each image shows, including a six-year-old's rash, a burn with a tattoo and part of a face in the frame, and a mole photo sent to a dermatologist by WhatsApp.
Can we use ChatGPT or Claude for a clinical photo register if we remove the names?
Not safely, because the rest of each row still points to one patient. 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 81-year-old with a skin tear after a fall, a 36-year-old whose burn photo went to an insurer with his employer's HR manager on copy, and a child in a group of seven staff can each be recognised by anyone who knows them. The conditions 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 we upload a clinical photo register spreadsheet to ChatGPT or Claude?
An uploaded register leaves the practice as a whole file. The Policy draft tab, any hidden columns, comments, tracked changes and file properties can travel with the Photo audit tab. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Besides the patients, the file would carry which doctors and nurses used personal phones, which consents were never written down and the COUNTIF formula that counts them.
How can a medical practice use AI for a clinical photo register 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. A draft photography policy is produced from the practice's own register and records, and a person checks the counts against the patient records. A doctor then approves the policy before it is used, and the camera roll copies still have to be deleted by the people who hold them.
Are clinical photos part of the patient's health record?
The RACGP says clinical photos taken for the purposes of patient management are part of the person's health record, even if they only exist in electronic form. It says they should be treated like any other personal health data and are subject to the same conditions for collection, disclosure and storage under state, territory and federal laws. The RACGP page was last updated on 29 June 2026.
Should a GP take clinical photos on a personal phone?
The RACGP says a clinician's own phone might not be as secure as practice devices and is more likely to be lost, stolen or accessed by other people, and that photos can be forgotten on the device. Its key points are to store photos in the patient's health record, always delete them from the device and avoid third-party storage and automated back-ups, under a practice policy.
What consent does a practice need to photograph a patient?
Good medical practice describes informed consent as a voluntary decision made with knowledge and understanding of the benefits and risks, and asks doctors to use consent processes for releasing and exchanging health information. Under APP 3.3 an organisation generally needs the person's consent to collect sensitive information, unless an exception such as a permitted health situation applies. The RACGP suggests a practice policy covering consent.

See Dominion answering from your own documents.

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

Book a demo