Workers compensation certificates and AI privacy: what a certificate of capacity gives away in ChatGPT or Claude.

A workers compensation certificate is written for readers outside the practice: the employer, the insurer and often a rehabilitation provider. GPs fill in many of them, the wording matters, and the notes behind them hold far more than the employer should see, which is why the draft is often pasted into ChatGPT or Claude to tidy up. This page marks every item in a routine certificate and the notes beside it, and shows how the same certificate is prepared on Ermos inside the practice.

The workers compensation certificate

Completed by the treating GP after an injured worker is seen, then given to the worker, who passes it to the employer and the insurer. It is updated at each review while the worker recovers.

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

Workers compensation certificate · Review 4 November 2026

Worker:

Employer and claim:

Injury.

Capacity.

Treatment.

GP notes (kept off the certificate):

An invented certificate and notes. The employer, insurer and claim number are fictional; mobile and provider numbers are masked. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    The worker’s name, date of birth and mobile

    Identifies the patient and how to reach her. Personal information, held by the practice as health information because it was collected in providing a health service. Privacy Act guide.

  2. 2
    Employer, insurer and claim number

    Ties the patient to her workplace and an open claim. A claim number is a key that unlocks a whole insurer file.

  3. 3
    The injury and the imaging result

    The diagnosis and an MRI finding. Health information is sensitive information under the Privacy Act. Patient data guide.

  4. 4
    The capacity assessment

    What she can and cannot do at work, which decides her hours and pay while she recovers. Patient data guide.

  5. 5
    Treatment and medication

    A prescription and a treatment plan against a named worker. Patient data guide.

  6. 6
    Unrelated mental health and claim history

    An anxiety diagnosis, an antidepressant and an earlier injury claim. None of it belongs on the certificate, and it is the most damaging thing on the page if it reaches the employer or insurer by another route. Patient data guide.

  7. 7
    A workplace conflict told in confidence

    An allegation about a named supervisor and a fear of retaliation, with an explicit request that the employer is not told. Good medical practice 4.4 asks doctors to hold this in confidence. Patient data guide.

  8. 8
    The rehabilitation provider

    A third party involved in the claim, with a direct mobile number.

  9. 9
    The certifying GP

    A register name and a masked provider number, linked here to this claim.

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 general practitioner (GP) practices. See how this works in practice: Certificates and reports on Ermos.

Common questions

Do ChatGPT and Claude use the information in a workers compensation certificate?
Yes, a pasted certificate and its notes are 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 draft carries an open claim number, an MRI result, an antidepressant, an earlier accident claim and an allegation about a named supervisor.
Is it safe to paste a workers compensation certificate into ChatGPT or Claude if we remove the worker's name?
No, because the claim number alone leads straight back to her. 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 employer, an insurer, an injury date and a claim number identify one worker to anyone at that workplace. 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 a GP uses their own ChatGPT or Claude account to word a workers compensation certificate?
The practice cannot see what was sent from a personal account. 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. A GP who pastes these notes to tidy the capacity wording sends the worker's mental health history and her complaint about her supervisor outside the practice's own records, so the practice could not say where that information went.
How can a GP use AI for a workers compensation certificate 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 certificate is drafted from the consultation notes, and only what the claim needs goes onto it. The GP checks it against the notes and signs it.
The employer and insurer will see the certificate anyway. What is the risk?
The certificate is meant for them; the notes behind it are not. A draft pasted into an AI tool usually carries the notes too, including unrelated history and things the patient asked the GP to keep private, and sends them to a recipient the patient never agreed to.
Does a worker's consent to the claim cover an AI tool?
A worker's authority for a claim generally covers the people involved in the claim, such as the employer, the insurer and the rehabilitation provider. Under APP 6.1, using or disclosing health information for another purpose needs consent or an APP 6.2 exception, and an AI provider is outside the claim.
What should a GP leave out of a certificate?
Anything the reader does not need to manage the worker's capacity and treatment for the injury. Unrelated diagnoses and confidences belong in the notes, which stay in the practice.

See Dominion answering from your own documents.

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

Book a demo