Complaint responses and AI confidentiality: what a reply to a client gives away in ChatGPT or Claude.

A complaint response in a brokerage is written under two regimes: ASIC RG 271 on internal dispute resolution, and section 9 of the Code of Practice, with AFCA and the Code Compliance Committee as the possible next readers. It has to be fair, complete and defensible, which is why it gets drafted in ChatGPT or Claude. It carries the whole story: what went wrong, what the file shows, what the insurer said, and what the brokerage will offer. This page marks every item in a complaint response that the Code and the Privacy Act bear on, and shows how the same response is drafted on Ermos inside the brokerage.

The complaint response

Sent within the timeframe RG 271 and section 9 of the Code of Practice require after a client complains, and provided to AFCA or the Code Compliance Committee if escalated.

Pacific Harbour Insurance Brokers

Ms Lena Marchetti¹
Marchetti Dental Pty Ltd¹
By email

Re: Your complaint of 20 October 2026, cyber claim

Dear Lena,

Thank you for your email and for the call on Thursday. I have reviewed the file with our compliance manager and set out our response.

Your complaint. You say that we should have arranged cyber insurance after the 2025 phishing incident, that the practice has now suffered a second loss of $61,000 through a fraudulent supplier payment, that this loss is uninsured, and that we are responsible.²

What our file shows. Our needs analysis of 15 October 2026 identified cyber as a gap and recorded that you wanted cover and Marco considered it unnecessary. Our email of 16 October recommended a quote from Anchor Underwriting for $1 million of cover at $4,200 and asked for instructions. No instructions were received before the loss on 19 October. We accept that we did not follow up by phone, which our process requires within five business days, and I apologise for that.³

The 2025 incident. After the 2025 loss we recommended cyber cover in our renewal letter of 10 November 2025. Marco declined it in writing on 14 November 2025, citing cost. A copy is enclosed.⁴

Our position. We do not consider the brokerage responsible for the uninsured loss, because cover was recommended twice and declined or not instructed. We acknowledge the follow-up failure. Without admission, we offer to waive our broker fees for the coming renewal ($650) and to arrange the Anchor cyber cover at no fee if you wish to proceed now.⁵

I am also aware from our conversation that the loss has caused difficulty between you and Marco, and that the practice’s bank has frozen the account pending its own investigation. If a letter from us confirming the sequence of events would assist with the bank, we will provide one.⁶

If you are not satisfied with this response, you may refer the matter to the Australian Financial Complaints Authority (1800 931 678), and you may report an alleged breach of the Code of Practice to the Insurance Brokers Code Compliance Committee.⁷

Priya Desai⁸
Account Manager⁸

An invented letter. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the brokerage

The complainant and the practice

Named, with the fact of a complaint. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	Transmitted to the provider, generally offshore on personal plans
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The allegation and a second fraud loss

What the client says, and that the practice has been defrauded twice, with the amount. Commercial information, and a security fact no business wants known.

Frontier Models (ChatGPT, Claude)	A second fraud loss transmitted; on consumer tiers it can train future models
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The brokerage’s file and its apology

The documents that defend the brokerage, and an admission that its follow-up process was not followed. The sentence a professional indemnity insurer would read first.

Frontier Models (ChatGPT, Claude)	Your own apology now on a third party’s servers
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

A director’s written refusal

That a named director declined the cover in writing. The decisive document, attributed to an individual. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A named director’s refusal transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The position and the offer

What the brokerage will concede and what it will not. The negotiating position in a dispute that may reach AFCA.

Frontier Models (ChatGPT, Claude)	The negotiating position transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

A dispute between the partners and a frozen account

That the loss has damaged the partners’ relationship and the bank has frozen the practice’s account. Confidential, and nothing to do with the complaint’s merits. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A partners’ dispute and a frozen account transmitted
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The escalation path

AFCA and the Code Compliance Committee. Operational.

Frontier Models (ChatGPT, Claude)	Transmitted and kept on the provider’s terms
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

The broker

The staff member whose follow-up failure is admitted, named.

Frontier Models (ChatGPT, Claude)	Transmitted, and the leak is traceable to you
Ermos Edge	Stays on your instance in Australia; leaves only inside a permitted excerpt, per question, to the named provider
Ermos Dominion	Stays on the unit in your office

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 Edge	Ermos Dominion
Where the document is processed	Provider servers, generally offshore on personal plans	Your instance in an Australian data centre; answers generated in Australia	The unit in your office
Who can read it	Provider staff and systems, under policy	Your staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothing	Your staff only
Is it kept	On the provider’s terms	Kept on your instance; the provider is barred from retaining it	Stays on the unit
Can it train a model	On consumer tiers, yes	No, by contract	Only your own unit learns from it, on your hardware

Part of AI data leakage in insurance brokerages. See how this works in practice: Client correspondence on Ermos.

Common questions

Do ChatGPT and Claude use the information in a complaint response?

Both receive the full response on the AI company's servers once it is pasted. 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 complaint response puts the brokerage's apology, its offer and a frozen bank account in one paste.

Is it safe to paste a complaint response into ChatGPT or Claude?

No, because the letter is the brokerage's full account of one dispute. It names the complainant, records the director's written refusal of cover and a second fraud loss, and mentions a dispute between the partners. Each detail points to one client, and the dated sequence of emails and calls makes the file easy to match. 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.

Does a ChatGPT Business or Claude Team plan make it safe to use for a complaint response?

Only in part. 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. A response that admits a missed follow-up, makes an offer and states the brokerage's position still reaches the AI company, whichever plan the brokerage buys. Our comparisons with ChatGPT Enterprise and Claude weigh the trade-offs.

How can a brokerage use AI for a complaint response 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 response is drafted from the needs analysis, the renewal letter, the emails and the policy wording. The broker and the compliance manager check a chronology against the documents it cites before the broker signs.

AFCA or the Code Compliance Committee may see the letter anyway. Why does the draft matter?

Both are bodies the law and the Code provide for, bound by their own rules. A frontier model provider is a company with servers, for the major ones outside Australia, retaining the draft on its own terms, with the brokerage's admission and the partners' dispute in it.

What does section 9 of the Code of Practice require?

That subscribers have a complaints process, handle complaints fairly and in set timeframes, and tell clients how to escalate to AFCA. RG 271 is ASIC's guide to internal dispute resolution. The standards highlighted in it are enforceable, and they include a response within 30 calendar days for a standard complaint. Neither says anything about drafting tools; both require the brokerage to keep the record.

How is a complaint response drafted on Ermos?

From the needs analysis, the renewal letter, the emails and the policy wording on your own unit, with a chronology the broker and the compliance manager check against the documents it cites. Nothing in the file leaves the brokerage to produce 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

Keep reading

AI data leakage in insurance brokerages AI for insurance brokers The Privacy Act, the APPs and AI All professions