Dental Post-Operative Instruction Messages: Use case and AI Confidentiality

After an extraction under sedation the practice texts the patient the aftercare instructions, checks on them the next morning and answers whatever comes back. The thread fills quickly, so staff paste it into ChatGPT or Claude to write a calm reply, to turn the standard sheet into plain English or to summarise two days of messages for the dentist. Post-operative instructions are generic. The follow-up messages around them are the opposite: what was done, which medicines the patient takes, a numb lip the morning after, a photo of a swollen face and a request about his employer. This page marks each item in a routine thread against the Privacy Act and the Dental Board’s Code of conduct.

The post-operative instructions and follow-up messages

Sent from the practice’s SMS tool on the day of a surgical procedure and for a few days after. Section 8.3 of the Code of conduct asks practitioners to keep records that are sufficient for continuity of care and to hold them securely, including electronic records, and section 4.5 asks them to explain promptly and fully when something has gone wrong.

Thu 12 Nov, 11.05am Tallowwood Dental:

Thu 12 Nov, 11.06am Tallowwood Dental:

Thu 12 Nov, 11.07am Tallowwood Dental:

Thu 12 Nov, 6.40pm Marta Wrona:

Thu 12 Nov, 6.52pm Sam (dental assistant): Thanks Marta. Tomasz told us on 27 October that we can talk to you. It is common after sedation and Dr Lee has it in his notes. Nothing to be embarrassed about.

Fri 13 Nov, 9.10am Tallowwood Dental: Good morning Tomasz. How are you today? Please reply with your pain from 0 to 10 and anything that worries you.

Fri 13 Nov, 9.34am Tomasz Wrona:

Fri 13 Nov, 9.50am Dr Jonathan Lee:

Fri 13 Nov, 10.02am Tomasz Wrona:

Fri 13 Nov, 10.15am Josh (reception):

Sat 14 Nov, 8.20am Marta Wrona:

Sat 14 Nov, 8.41am Dr Jonathan Lee: Swelling is usually at its worst on day two or three. Keep up the cold packs and the rinses. If his temperature goes over 38.5, the swelling spreads under the jaw or he has trouble swallowing or breathing, go to an emergency department and ring me on the way.

Sat 14 Nov, 8.55am Tomasz Wrona:

An invented thread. The practice, staff, patient, his wife and his manager are fictional; the card digits are placeholders. The aftercare wording is an example and is no substitute for a dentist’s instructions. Hover or tap any highlighted item for the analysis; the full list is below.

Post-operative message drafter

An Agent+ that prepares the messages a patient receives after surgery. It starts from the practice's own approved instructions and the procedure record, drafts the day-of and follow-up texts, and prepares a draft reply when a family member or the patient asks something.

What it does

  1. Reads the procedure record and approved instructions
  2. Drafts the day-of and follow-up messages
  3. Prepares a draft reply for patient questions

What stays with your people. The dentist or nurse reviews and sends the messages. Any new symptom is handled by a clinician.

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 patient and the appointment

    A named patient, the day, the dentist, the fact of sedation and who took him home. Health information under the Privacy Act, because it is collected in providing a health service, held by a practice the Act covers whatever its turnover. Patient data guide.

  2. 2
    What was done

    One tooth removed and the roots of another left in place because of the nerve. The operative detail of a named patient, and health information. Section 8.3 of the Code of conduct asks practitioners to hold health records securely, including electronic records. Patient data guide.

  3. 3
    The standard aftercare instructions

    Gauze, rinsing, food, alcohol and the 24 hours after sedation. The practice’s standard sheet, sent to every patient who has this procedure. It contains nothing about this patient, and it is the one part of the thread that is reasonable to ask any tool to reword, provided a dentist checks the result.

  4. 4
    Medicines and a request about opioids

    Pain relief, an antidepressant, a CPAP machine and a note that he asked for no opioid. Health information that says more about him than the surgery does. Australian Privacy Principle 11.1 asks an entity to take reasonable steps to protect personal information from unauthorised access and disclosure. Patient data guide.

  5. 5
    His wife’s message

    A third person describing a panic on waking and his embarrassment about it. The practice answers her because he said on 27 October that it could. Section 3.3 of the Code asks practitioners to seek informed consent before disclosing information, and that consent covered his wife. It did not cover an AI provider. Privacy Act guide.

  6. 6
    The numb lip and the beers

    A symptom that may be a nerve injury, and an admission about alcohol the night after sedation. Health information volunteered because he trusts the practice, and the first record of a possible complication. Patient data guide.

  7. 7
    The dentist’s reply

    An explanation of what the numbness might be, what was done to lower the risk, and a review booking. Section 4.5 of the Code asks practitioners to explain to the patient as promptly and fully as possible what has gone wrong and how it might affect them. The wording is clinical advice, which Ahpra’s guidance on AI says stays the practitioner’s responsibility however it was drafted.

  8. 8
    The certificate and the employer

    His manager’s name, his roster, and what he does not want his employer to know. The practice gives the certificate to him and leaves the procedure out. Australian Privacy Principle 6.1 limits disclosure for another purpose unless the person consented or an exception applies. Privacy Act guide.

  9. 9
    The photo and the temperature

    A photograph of his face, a temperature and how far he can open his mouth. Health information, and an image that identifies him. Section 3.3 of the Code asks practitioners not to transmit or share a person’s information or images without first getting written and informed consent. Patient data guide.

  10. 10
    The balance and the card

    What he still owes, how he will pay it and the last digits of his card. Personal financial information, which is outside the Privacy Act’s list of sensitive information and is still his. Privacy Act guide.

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 dental practices. See how this works in practice: Recalls on Ermos.

Common questions

Do ChatGPT and Claude use the information in post-operative instructions and follow-up messages?
A pasted message thread goes in full to the AI company's servers and is 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 thread links Tomasz Wrona to sertraline and a CPAP machine. It also records two beers on the night of his surgery, a numb lip the next morning and a photo of his swollen cheek.
Is it safe to paste post-operative follow-up messages into ChatGPT or Claude if we remove the names?
No, apart from the standard aftercare instructions, which hold nothing about the patient and are reasonable to reword if a dentist checks the result. 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. Without names, the thread still holds a photo of his face, a station manager, a night roster from Saturday and a wife answering for him. His sedation, medicines and possible nerve injury are health information, which the Privacy Act lists as sensitive information. He agreed the practice could talk to his wife, and that consent did not extend to an AI company. 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.
What if staff use their own ChatGPT or Claude account for post-operative follow-up messages?
Then the practice cannot see or manage what was sent. On a personal plan the user cannot choose where conversations are stored or processed. Some business plans let an organisation choose to store data at rest in Australia (OpenAI), which still leaves the content with the AI company. Both companies can also review content that is flagged for safety or policy reasons. This thread runs from Thursday into Saturday morning, and a reply drafted out of hours in a personal account leaves no trace at the practice. Nobody there could later say whether the photo of his face, his sertraline or his wish to keep the sedation from his employer was sent.
How can a dental practice use AI for post-operative follow-up messages without sending them 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. Replies are drafted from the patient's record and the practice's own aftercare sheet. Dr Lee reads and owns every word of clinical advice, and reception checks the two payments of $925 and the certificate wording.
Are text messages with a patient part of the dental record?
They can be. Section 8.3 of the Code of conduct asks practitioners to keep accurate, up to date records that are sufficient for continuity of care, to make them at the time of events or as soon as possible afterwards, and to hold them securely, including electronic records. A message in which a patient reports a numb lip the morning after surgery is clinical information that the next person treating him needs to see.
Can a dental practice discuss a patient's recovery with a family member?
Section 3.3 of the Code of conduct asks practitioners to respect a patient's confidentiality and privacy by seeking informed consent before disclosing information. In practice that means the patient tells the practice who it may speak to, and the practice records it. Consent given for a wife to receive updates does not extend to anyone else, such as an employer, and what goes on a medical certificate is a separate question.
What should a dentist do when a patient reports a problem after surgery?
Where a patient has been harmed by healthcare, section 4.5 of the Code of conduct asks practitioners to recognise what has happened, act to rectify the problem where possible, and explain to the patient as promptly and fully as possible what has gone wrong, how it happened and how it might affect them. It also asks them to document the event and to seek advice if unsure of their obligations.

See Dominion answering from your own documents.

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

Book a demo