Mental health treatment plans and AI privacy: what a GP mental health plan gives away in ChatGPT or Claude.

A GP mental health treatment plan is among the most personal documents a practice produces. It records symptoms, scores, a provisional diagnosis, an assessment of risk, substance use, goals and who the patient does and does not want told. It is long, structured and repetitive to write, which is why doctors paste their notes into ChatGPT or Claude to fill the template. This page marks every item in a routine plan that the Privacy Act, Good medical practice and the NSW health records law bear on, and shows how the same plan is drafted on Ermos from the consultation inside the practice.

The GP mental health treatment plan

Prepared by a GP after assessing a patient with a mental disorder, and used to refer the patient to a psychologist or other allied health provider. Medicare item numbers 2715 and 2717 are the plan items for GPs with mental health skills training.

GP Mental Health Treatment Plan
Kestrel Lane Medical Practice · · 19 October 2026

Patient
Presentation
Assessment
Risk
Substances and physical health
Goals and plan
Work and family
Consent and sharing
Billing and review

An invented treatment plan. The Medicare number is a placeholder. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    Name, date of birth and Medicare number

    The identifiers that tie the plan to one person. APP 9.2 limits when an organisation may use or disclose a government related identifier such as a Medicare number. Privacy Act guide.

  2. 2
    The presentation

    Symptoms and their link to a workplace event. Health information is sensitive information under the Privacy Act. Patient data guide.

  3. 3
    Screening scores and a provisional diagnosis

    K10, PHQ-9 and GAD-7 results and a working diagnosis. The core clinical content, and the part an insurer or employer would most like to see. Patient data guide.

  4. 4
    The risk assessment

    Self-harm thoughts, protective factors and a safety plan. The most sensitive paragraph in the record, held in confidence under Good medical practice 4.4. Patient data guide.

  5. 5
    Alcohol, cannabis and tests ordered

    Substance use and the investigations the doctor ordered. Health information, and detail the patient told the doctor in confidence. Patient data guide.

  6. 6
    Goals, treatment and the referral

    The plan agreed with the patient, a named psychologist and a medication change. Clinical judgement that a doctor must check, and information a psychologist will also hold. Patient data guide.

  7. 7
    Employer and family circumstances

    A named employer, a plan to resign and who has not been told. The sort of detail that would hurt if it travelled. Privacy Act guide.

  8. 8
    Consent and limits on sharing

    Who may receive the plan, who may not, and a refusal of AI scribing. The patient’s expressed wishes, which the practice must follow. HPP 11 of the NSW health records law limits disclosure of health information. Patient data guide.

  9. 9
    The billing item and the follow-up date

    Item 2715 identifies the service as a GP mental health treatment plan, and the date shows when the patient returns.

  10. 10
    The GP

    Ties the plan to a named doctor bound by Good medical practice.

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

Common questions

Do ChatGPT and Claude use the information in a GP mental health treatment plan?
Yes, a pasted treatment plan 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 plan records K10, PHQ-9 and GAD-7 scores, a provisional diagnosis of major depressive disorder, a risk assessment covering thoughts of self-harm, and who must not be told.
Is it safe to paste a GP mental health treatment plan into ChatGPT or Claude?
No, and tidying the plan first changes little. 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. Mental health information is health information, which is sensitive information under the Privacy Act. 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 named employer, a plan to resign and a partner who knows still point to one person after the name goes.
Can ChatGPT or Claude remember the details in a GP mental health treatment plan later?
They can. 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). Those details could follow the user into a later chat about a different patient. This plan holds screening scores, a safety plan and substance use, and the patient asked that nothing go to parents or employer.
How can a medical practice use AI for a GP mental health treatment plan 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 plan is drafted from the consultation record and the practice's own plan template. The GP assesses the patient, checks every entry and signs it.
Is mental health information treated differently from other health information?
Under the Privacy Act, mental health information is health information, which is sensitive information, so the same stricter rules apply as for any clinical detail. It is also the information patients most expect to be held in confidence. Good medical practice 4.4 asks doctors and their staff to hold patient information in confidence unless release is required or permitted by law.
A patient said no copy to family or employer. How does that interact with AI?
A patient's wishes on sharing are part of the record. HPP 11 of the NSW Health Records and Information Privacy Act 2002 limits disclosure of health information, and APP 6 does the same federally. Pasting the plan into an outside AI tool is a further recipient the patient never agreed to, so the instruction is easier to keep if the drafting stays in the practice.
Can AI help draft a treatment plan?
It can help with structure and wording, but the doctor assesses the patient and owns the plan. Ahpra says the practitioner remains responsible and must apply human judgment to AI output. The privacy question is where the drafting happens, which is why the plan should be drafted on infrastructure the practice controls.

See Dominion answering from your own documents.

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

Book a demo