Dental Staff Performance Reviews: Use case and AI Confidentiality

A performance review is a document a practice manager writes alone, late in the day, and wants to get right, so it is a natural thing to paste into ChatGPT or Claude to soften the wording or turn rough notes into a finished form. It records the employee’s ratings, the incident that went wrong, often the name of a patient who complained, what the employee has told the manager about their health and family, and the pay outcome. This page marks every item in a routine six-month review of a dental receptionist that the Privacy Act, NSW health records law and the Dental Board’s Code of conduct bear on, and shows how the same form is drafted on Ermos inside the practice.

The staff performance review

Completed at the end of a probation period or each year. The review is about an employee, but in a dental practice it usually mentions patients, and the Code of conduct (section 3.3) asks practitioners to ensure all staff are aware of the need to respect patients’ confidentiality and privacy.

Six-month probation review
Tallowwood Dental · Friday 13 November 2026, 2.00pm · Cheryl Dunmore, Practice Manager, with Dr Anika Rao for item 6

Employee
1. Ratings (1 to 5)
2. Incident
3. Strength
4. Wellbeing
5. Hours
6. Outcome
7. Goals to 15 January 2027
Sign-off

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

Performance review preparer

An Agent+ that prepares a staff review in the practice's format. It works from the position description, the manager's notes and earlier reviews, drafts the ratings and comments sections, and flags patient names or personal health information to keep out of the form.

What it does

  1. Reads the position description and manager's notes
  2. Pre-fills the practice's review form
  3. Flags patient names and personal details to remove

What stays with your people. The principal or practice manager completes the review and decides outcomes, with the employee. Employment questions go to the practice's adviser.

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 employee

    A named employee, his role, hours, start date and mobile. Personal information about him. An employer’s act or practice that is directly related to a current or former employment relationship and to an employee record it holds is exempt from the Privacy Act under section 7B(3), and the NSW health records Act leaves employee record information held by a private sector person out of its definition of personal information, so those laws may not be the main constraint on this line. His trust in the practice still is. Privacy Act guide.

  2. 2
    The ratings

    Scores for communication, accuracy, teamwork, reliability and systems. Performance is one of the examples of employee record information in section 6(1) of the Privacy Act. A leaked score is also a personal judgment about a named person, and the AI may soften it into something the manager did not decide.

  3. 3
    The incident and the patient who complained

    A named patient, her treatment, the fee dispute and the credit offered. This is information about her rather than about the employee, and it sits inside a document about his employment. Do not assume the employee record exemption reaches what is said about a patient. It may also be health information, because section 6 of the Health Records and Information Privacy Act 2002 includes other personal information collected in providing a health service, and the Health Privacy Principles (HPP 5) ask for reasonable safeguards and, where information must be given to a service provider, everything reasonably within the practice’s power to prevent unauthorised use or disclosure. Patient data guide.

  4. 4
    A patient’s family and a patient’s anxiety

    A patient and an email from his wife, which says he is very anxious and was having sedation. Health information about Tomasz Wrona and personal information about Marta. Section 3.3 of the Code asks practitioners to respect patients’ confidentiality and to ensure staff are aware of that need. Patient data guide.

  5. 5
    The employee’s health

    That he is anxious and seeing his GP. Health information about the employee is named as an example of an employee record in section 6(1) of the Privacy Act. It is also the most personal thing in the form, and a manager’s note of a confidence.

  6. 6
    His mother’s stroke and his hours

    A family member’s medical event and a standing arrangement about working hours. His hours and leave are employee record examples. His mother’s health is information about someone who is not an employee, and it is unclear from the text whether the exemption reaches it, so treat it as outside.

  7. 7
    The probation outcome and pay review

    The decision and the pay review date. Salary and wages, and the engagement of the employee, are examples of employee record information in section 6(1). It also tells a reader the manager’s recommendation before the owner has decided.

  8. 8
    Goals and the script

    Training goals, a deadline and a suggested sentence for reception. Generic role content that would apply to any receptionist.

  9. 9
    The employee’s comment and the signatures

    What the employee said when he signed, and the record that he, the manager and the principal agreed. A signed personnel record, with his own words attached.

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.

Common questions

Do ChatGPT and Claude learn from a staff performance review?
Yes, a pasted review 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. A probation review of a receptionist can hold his ratings, a patient's name and fee complaint, his visits to a GP for anxiety and his mother's stroke.
Is it safe to paste a staff performance review into ChatGPT or Claude if we remove the employee's name?
No, because the review still describes one person. 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. A receptionist on four days a week who started in May, leaves at 3pm on Thursdays and has a mother recovering from a stroke is easy to recognise inside a small practice. The Privacy Act's employee records exemption (section 7B(3)) may cover the practice's handling of his own record, but nobody should assume it reaches what the review says about a patient. 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 a staff performance review?
Then the practice cannot see or manage what was sent. A manager who drafts a review late in the day on a personal login may paste the ratings, the patient complaint and the employee's health notes without anyone knowing. 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. The review would also sit in an account the practice does not control, tied to the manager's own email address.
How can a dental practice use AI for a staff performance review 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 review is drafted from the manager's notes and the practice's own review template. The practice manager checks every rating and quote, leaves patient names out, and the principal decides the pay outcome.
Does the Privacy Act cover a staff performance review?
Often only partly. Section 7B(3) of the Privacy Act exempts an employer's acts or practices directly related to a current or former employment relationship and an employee record it holds about that individual, and performance is listed as an example of employee record information. The exemption does not obviously reach a patient's details quoted in the review, so treat those as protected.
What if a review names a patient?
The patient's name and the dispute are information about the patient. In NSW a dental practice is a health service provider under the Health Records and Information Privacy Act 2002, and HPP 5 asks for reasonable safeguards and, where information goes to a service provider, everything reasonably within the practice's power to prevent unauthorised use or disclosure. The simplest course is to leave patient names out of HR documents.
What does the Dental Board expect staff to understand about confidentiality?
Section 3.3 of the Code of conduct asks practitioners to ensure that all staff are aware of the need to respect the confidentiality and privacy of patients, and to avoid discussing patients in a non-professional context. Ahpra's guidance on using AI in healthcare, which is written for clinical use, adds that the practitioner remains responsible for safe and quality care and should understand whether patient data is used to train the model and where it is stored.

See Dominion answering from your own documents.

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

Book a demo