Lender submission emails and AI data leakage: what a broker’s application email gives away in ChatGPT or Claude.

The submission email is the broker’s pitch to a lender’s business development manager: who the applicants are, what they want, why the numbers work and what the policy exception is. It is written fast, often after a decline elsewhere, and brokers paste the draft into ChatGPT or Claude to tighten it. It also carries the broker’s candid asides about the clients and the attachments about to follow. This page marks every item in a routine lender submission email that the National Consumer Credit Protection Act and the Privacy Act bear on, and shows how the same email is drafted on Ermos inside the brokerage.

The lender submission email

Sent when the application is ready to lodge, often with a policy exception request. It is part of the information provided to the credit provider, which ASIC’s RG 273.165 lists among the records a broker is expected to keep.

From: Anita Menon   To: Jonas Whitfield, Business Development Manager, Southgate Bank
Wed 21 Oct 2026, 4.42pm   Subject: New submission: Whitlock D & S, $868,000 purchase

Hi Jonas,

Submission attached for : .

Income.

Deposit.

Credit.

Attached.

Happy to talk it through if useful.

An invented email. The lender, its manager and the phone number are invented; the licence value is a placeholder. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the brokerage

  1. 1
    The applicants

    Named. Everything below is identifiable. Privacy Act guide.

  2. 2
    The property and the loan

    The address, price, LVR and structure. The deal in one sentence. Privacy Act guide.

  3. 3
    Income and the assessed gross

    Employers, base pay, overtime and shading. Financial information supplied to the credit provider after the verification section 117 requires. Privacy Act guide.

  4. 4
    The deposit and a gift

    Savings, the gift, the donor’s name and the date the money landed. A third person’s finances in a document that will reach the lender. Privacy Act guide.

  5. 5
    Credit conduct

    A late payment on a named card, its length and the explanation, and a count of recent enquiries. Credit information about an individual; Part IIIA of the Privacy Act regulates credit reporting information and who may obtain it (section 24). Privacy Act guide.

  6. 6
    The attachments list

    A list of document names. Low exposure alone, and a precise inventory of what the brokerage holds on the clients.

  7. 7
    The broker’s candid aside

    A frank assessment of the clients and of their employer’s position. Written for one reader at the lender, never for a general-purpose tool.

  8. 8
    An earlier decline and its reason

    That another lender said no and why. Commercial information about the application and the other lender’s policy.

  9. 9
    Broker, licence and phone

    Public register details and a business number. The licensee that owes the duty in section 158LA, and the sender a scammer would impersonate.

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 AI data leakage in mortgage brokerages. See how this works in practice: Application packaging on Ermos.

Common questions

Do ChatGPT and Claude use the information in a lender submission email?
Yes, the draft 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. The email names the applicants, a $60,000 gift from a relative, a late payment on a Visa and the broker's candid aside about Daniel's nerves over his employer.
Is it safe to paste a lender submission email into ChatGPT or Claude?
No, because the email gathers the deal, the income and the credit history in one message. 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. Taking out the names still leaves the Glenmore Park address, the $868,000 loan, a $60,000 gift from the applicant's father and a late payment on a named card. 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 lender submission email?
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 email carries the clients' income, a relative's gift and the broker's candid aside, and all of it would reach the AI company before the lender has read the submission. Our comparisons with ChatGPT Enterprise and Claude weigh the trade-offs.
How can a mortgage brokerage use AI for a lender submission email 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 email is drafted from the client file and the brokerage's own submission template. The broker checks every figure against the file and decides what the lender should read before it goes out.
What does a broker owe the client when asking a lender for an exception?
The best interests duty in section 158LA or 158LE applies to the credit assistance as a whole, including how the application is presented. A licensee must also do all things necessary to ensure credit activities are engaged in efficiently, honestly and fairly (section 47(1)(a)). An exception request should be accurate and supported by the file, because the lender will rely on it.
Is pasting the submission into an AI tool a disclosure of the clients' information overseas?
If the tool's servers are outside Australia, APP 8.1 requires an APP entity (a brokerage outside the small business exemption) to take reasonable steps to ensure the overseas recipient does not breach the APPs, and in some circumstances the recipient's act is treated as the entity's own. The answer turns on the tool, the plan and the brokerage's terms, which is why the destination matters.
Why do lenders and aggregators care where the broker's emails are drafted?
Brokers send lenders information their clients gave them in confidence, and lenders and aggregators may set conditions on how broker systems handle it. Section 47 also requires a licensee to have adequate arrangements and systems to ensure compliance and adequate risk management systems. Where a draft is written is part of that picture.

See Dominion answering from your own documents.

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

Book a demo