Law Firm Medical Evidence Summaries: Use case and AI Confidentiality

A personal injury solicitor spends much of a file turning medical reports into a summary: diagnoses, history, prognosis, what each doctor said that the others did not. It is the sort of dense reading ChatGPT or Claude is offered for, and the summary is built from the most sensitive information a client has. It also carries the solicitor’s view of whether the claim clears the statutory threshold. This page marks every item in a summary of a fall at a shopping centre that rule 9 and the Privacy Act bear on, and shows how the same summary is built on Ermos inside the firm.

The medical evidence summary

Prepared once treating and independent reports are in, to advise the claimant and to brief counsel before a settlement conference.

Medical evidence summary
· 9 November 2026 ·

Claimant. .

Incident.

Diagnoses.

Previous history.

Psychiatric.

Family.

Sources.

Our assessment.

An invented summary. The claimant, doctors and centre are invented. Hover or tap any highlighted item for the analysis; the full list is below.

Medical evidence summary agent

An Agent+ that reads the medical reports and subpoenaed records on a claimant's file and drafts a summary in your firm's format. It builds a dated chronology of conditions and treatment, notes where reports differ, and holds the solicitor's own assessment apart for counsel's brief.

What it does

  1. Reads the treating, independent and subpoenaed medical records
  2. Builds a dated chronology of conditions and treatment
  3. Lists where reports differ or details are missing

What stays with your people. A solicitor checks the summary against the source reports and the partner settles it. The agent summarises; it gives no medical opinion.

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 firm

  1. 1
    The claimant, her birth date and her address

    A named individual with a date of birth, address and occupation, in a claim against a named centre. Personal information, and confidential to the client under rule 9. Rule 9 confidentiality guide.

  2. 2
    The incident and the allegation against the centre

    Where and when she fell and an allegation that a warning sign was missing. The core of the liability case, before it is pleaded. Legal privilege guide.

  3. 3
    The diagnoses and the impairment figure

    A fracture, a surgery, a permanent loss and an impairment percentage. Health information is sensitive information under the Privacy Act, which applies to a law practice only if the practice is covered, for example through turnover above $3 million. Privacy Act guide.

  4. 4
    Prior conditions

    A 2019 back history, smoking and hypertension, and back complaints before the fall. The other side’s best material on causation, in the claimant’s own GP records. Privacy Act guide.

  5. 5
    The psychiatric history and the medication

    A diagnosis, a mother’s death, a drug and a named psychologist. Among the most sensitive health information a client has, and it is in a summary written for a negotiation. Privacy Act guide.

  6. 6
    A spouse’s illness

    A named husband with a progressive illness and the claimant’s caring role. His health information, supplied by someone else, and a third party who is not the client. Patient data guide.

  7. 7
    Records produced on subpoena

    Radiology and GP records obtained by compulsory process. Paragraph 9A of Practice Note SC Gen 23 says material produced on subpoena must not be entered into any Gen AI program unless the practitioner is satisfied of the conditions it lists. Paragraph 9B allows summarising documents subject to 9A. The practice note governs proceedings in the Supreme Court of NSW, so check the rules of the court that will hear the claim. Legal privilege guide.

  8. 8
    The firm’s view of the threshold and the range

    That the case is borderline, that the key witness has not been told something and the settlement range. Privileged, and the paragraph the defendant’s insurer would most like to read. Legal privilege guide.

  9. 9
    The statement of the test

    A short statement that the Civil Liability Act measures severity against a most extreme case, and that the impairment figure is not the test. General law with no client in it.

  10. 10
    The solicitors

    Ties the summary to practitioners bound by rule 9 and supervised by a partner. Supervising AI 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 firm 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 matter file. 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 law firms.

Common questions

Do ChatGPT and Claude use the information in a medical evidence summary?
Everything pasted in 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 summary carries the claimant's psychiatric diagnosis and medication, her husband's Parkinson's disease and the settlement range of $190,000 to $240,000.
Is it safe to paste a medical evidence summary into ChatGPT or Claude?
No, because a summary is built from the most sensitive information a client has. 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. A part-time bookkeeper in Epping who fell at a shopping centre on 11 March 2025, with a husband who has Parkinson's disease, is easy to pick out once her name is removed. 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.
Does a ChatGPT Business or Claude Team plan make it safe for a medical evidence summary?
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. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. A business plan changes the default on training, but it does not change who receives the content. This summary also draws on radiology and GP records produced on subpoena, and paragraph 9A of Practice Note SC Gen 23 asks the practitioner to be satisfied of its conditions before such records are entered into any Gen AI program. The note applies to the Supreme Court of NSW. Our comparisons with ChatGPT Enterprise and Claude weigh the trade-offs.
How can a law firm use AI for a medical evidence summary 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 reports and subpoenaed records are summarised from the matter file, with each statement linked to the page it came from. The solicitor checks every statement against the report.
What threshold must a personal injury claimant clear for non-economic loss in NSW?
Section 16(1) of the Civil Liability Act 2002 (NSW) provides that no damages may be awarded for non-economic loss unless the severity of the loss is at least 15 per cent of a most extreme case. The comparison is made by the court. A doctor's whole person impairment percentage can be evidence of the injury, but it is not the statutory test, and the Act applies to the claims it covers rather than to every injury claim.
Can a solicitor summarise medical records with an AI tool?
Practice Note SC Gen 23 paragraph 9B lists summarising or reviewing documents and transcripts as a permitted use, subject to paragraph 9A. Paragraph 9A says material produced on subpoena must not be entered into any Gen AI program unless the practitioner is satisfied it stays in a controlled environment, is used only for the proceeding and is not used to train a model. The practice note applies to proceedings in the Supreme Court of NSW, so check the rules of the court that will hear the claim.
How is a medical summary built on Ermos?
The reports and subpoenaed records are indexed on your own network. The unit builds a timeline of diagnoses and treatment, notes where reports disagree and links each statement to the page it came from. The solicitor checks every statement against the report and decides what the summary concludes.

See Dominion answering from your own documents.

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

Book a demo