Mortgage Broker Data Breach Assessment Notes: Use case and AI Confidentiality

After a document pack goes to the wrong person, the director has to write up what happened, decide whether it is a notifiable breach and record why. It is the document everyone wants to get right, so it is the one most often pasted into ChatGPT or Claude for a second opinion on the wording. The record holds the affected clients, the stranger who received the pack, the employee who made the mistake and the reasoning for staying silent. This page marks every item in a routine privacy incident assessment that the Privacy Act and the National Consumer Credit Protection Act bear on, and shows how the same note is drafted on Ermos.

The data breach assessment note

Written as soon as the incident is contained, and completed inside the 30 days the Privacy Act allows for assessing a suspected eligible data breach (section 26WH(2)). Whether the incident is an eligible data breach turns on unauthorised access or disclosure and whether a reasonable person would conclude it is likely to result in serious harm (section 26WE(2)), subject to remedial action (section 26WF). A credit licensee also checks whether the breach is a reportable situation under section 50A of the National Consumer Credit Protection Act.

Privacy incident assessment: misdirected email
Harbour Ridge Finance Pty Ltd · Confidential · completed Wednesday 11 November 2026 by Elena Voss, Director

What happened. At 2.14pm on Monday 9 November, the supporting documents pack for went by email to .

What was in the pack.

How it came to light and what we did.

Assessment.

Decision.

Reportable situation.

Changes.

Elena Voss, Director

An invented incident note. The brokerage, clients, recipient and staff are invented, and no identifier is real. Hover or tap any highlighted item for the analysis; the full list is below.

Data breach assessment note drafter

An Agent+ that organises an incident into your assessment note template. It works from the incident report, the list of documents involved and your privacy procedure, and produces a dated timeline, an inventory of affected clients and information, and a list of questions the director still has to answer.

What it does

  1. Reads the incident report and the documents involved
  2. Builds a timeline and a list of affected clients
  3. Lists questions the director must still answer

What stays with your people. A director or privacy officer reaches and records every conclusion, including whether anyone is notified. The agent organises the facts only.

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 brokerage

  1. 1
    The affected clients

    Two named people and the suburb they live in, in a document that exists because their information went astray. Privacy Act guide.

  2. 2
    The stranger who received the pack

    Another client’s name and the reason their address came up. A third party’s personal information, and a fact about the brokerage’s own client list. Privacy Act guide.

  3. 3
    The employee who made the error

    A named staff member tied to a mistake. Personal information about her, and an internal matter the brokerage would handle with her directly rather than with a general-purpose tool.

  4. 4
    What the pack contained

    Identity documents, payslips, statements and credit reports. An APP entity must take reasonable steps to protect the personal information it holds (APP 11.1), and Part IIIC of the Privacy Act applies to an eligible data breach of that information (section 26WE(1)(a)). Privacy Act guide.

  5. 5
    Spending that reveals health and family

    Payments that point to a child’s health care, and a family transfer. Health information is sensitive information under section 6(1) of the Privacy Act, and one of the factors for judging harm is the sensitivity of the information (section 26WG). Privacy Act guide.

  6. 6
    The containment and the written confirmation

    The call, the deletion request and the recipient’s written confirmation. The evidence for remedial action: if the entity acts before the disclosure results in serious harm and a reasonable person would then conclude serious harm is not likely, it is not an eligible data breach (section 26WF).

  7. 7
    The harm assessment

    The reasoning for concluding serious harm is not likely. Section 26WG lists the matters to weigh: the kinds of information, their sensitivity, protections, who obtained them, the nature of the harm and anything else relevant. The conclusion is the director’s to reach and sign.

  8. 8
    The decision, the dates and the turnover

    The awareness date, the completion date and the decision not to notify. The entity must carry out a reasonable and expeditious assessment and take all reasonable steps to finish within 30 days after becoming aware (section 26WH(2)), and the statement and notification duties in sections 26WK and 26WL apply to an eligible data breach. Whether the Act applies at all depends on turnover and the exceptions in section 6D.

  9. 9
    The reportable situation check

    The question whether this is a significant breach of a core obligation. Section 50A treats a breach of a core obligation as a reportable situation if significant, and section 50B(4) requires a report within 30 days after the licensee first knows, or is reckless about whether, there are reasonable grounds to believe one has arisen.

  10. 10
    The controls introduced

    The brokerage’s own security changes. Generic, though it shows what was not in place before.

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 brokerage 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 client 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 mortgage brokerages. See how this works in practice: Documentation best practice on Ermos.

Common questions

Do ChatGPT and Claude use the information in a data breach assessment note?
A pasted note 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 note names the clients whose pack went astray, the client who received it and the loan processor who sent it, and it lists the licences, passport and credit reports in the pack.
Can we use ChatGPT or Claude for a data breach assessment note if we remove the names?
Not safely, because the incident still describes real people after the names go. 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. The Oran Park suburb, the date and time of the email, the contents of the pack and the payments to a paediatric speech pathologist would all remain. Health information is sensitive information under section 6(1) of 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.
Does a ChatGPT Business or Claude Team plan make it safe for a data breach assessment note?
Not on its own. A business plan changes the default on training, but it does not change who receives the content. 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. The note would send the harm reasoning and the decision not to notify the OAIC to the AI company before the director has signed it. Our comparisons with ChatGPT Enterprise and Claude weigh the trade-offs.
How can a mortgage brokerage use AI for a data breach assessment note 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 note is drafted from the incident record on the file and the brokerage's own assessment template. The director checks the dates, the harm reasoning and the reportable situation question, and signs it.
How long does a business have to assess a suspected data breach?
Section 26WH(2) of the Privacy Act requires an APP entity, or another entity covered by section 26WE(1), that suspects it may have had an eligible data breach to carry out a reasonable and expeditious assessment and to take all reasonable steps to complete it within 30 days after it becomes aware of the grounds for suspicion. If it then has reasonable grounds to believe there has been an eligible data breach, it must prepare a statement and give it to the Commissioner as soon as practicable.
What makes a data breach an eligible data breach?
Under section 26WE(2), there must be unauthorised access to, or unauthorised disclosure of, information, and a reasonable person would conclude it is likely to result in serious harm to an individual it relates to, or information is lost in circumstances where that is likely to follow. Section 26WF says it is not an eligible data breach if remedial action is taken before serious harm results and a reasonable person would then conclude serious harm is not likely.
What must a notification statement say?
Section 26WK(3) requires the statement to give the identity and contact details of the entity, a description of the eligible data breach, the kinds of information concerned and recommendations about the steps individuals should take. Section 26WL then requires the entity to take reasonable steps to notify the contents to each individual the information relates to or to the individuals at risk, or otherwise to publish the statement on its website and publicise it.

See Dominion answering from your own documents.

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

Book a demo