Review letters and AI data leakage: what an annual review gives away in ChatGPT or Claude.

The annual review letter is the most routine document in an advice practice and the one most often handed to ChatGPT or Claude to be made warmer. It carries a year of the client’s financial life: balances, returns, what changed, what it cost, and what the adviser recommends next, alongside the fee consent the Corporations Act requires for ongoing arrangements. This page marks every item in a standard review letter that the Code of Ethics and the Privacy Act protect, and shows how the same letter is produced on Ermos from the client file inside the practice.

The annual review letter

Sent after the annual review meeting, with the ongoing fee consent form, to confirm the year’s position and the recommendations for the next.

Harbourview Wealth Partners

Tom and Lucy Harrington¹
9 Seaview Road, Freshwater NSW 2096¹

Dear Tom and Lucy,

Thank you for your time on Wednesday. This letter confirms where things stand after the review.

Your position. Tom’s superannuation is $438,200 (Southern Cross Super), up 6.4% for the year. Lucy’s consolidated balance is $198,700. The offset account holds $94,000 against the home loan balance of $598,000. The Tesla shareholding is valued at $31,400, down 17%.²

Changes. As we discussed, Tom’s role is expected to end in December. Lucy’s income protection application with Meridian Life has been accepted with an exclusion for thyroid-related conditions, which we recommend you accept.³

Our recommendations for the year. Pause the salary sacrifice increase (see the Record of Advice enclosed). Direct the redundancy payment to offset. Review the Tesla holding in January. Update both wills, for which we have referred Lucy to Carrick Doyle Lawyers.⁴

Fees. Our ongoing advice fee for the coming year is $4,400 including GST, deducted from Tom’s superannuation. The enclosed consent form sets out the services included. Meridian Life paid the practice a commission of $4,860 on Tom’s policy during the year.⁵

Lucy, I have also enclosed the information you asked for about the estate planning referral, as you requested, in a separate envelope.⁶

Please sign and return the fee consent within 30 days so that the arrangement continues without interruption.⁷

Kind regards,
Nadia Reyes⁸
Senior Adviser, Harbourview Wealth Partners⁸

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

What leaves the practice

The clients and their home

Named, at a residential address, with the home loan balance elsewhere in the letter. 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

Balances, returns and the loan

A household’s complete position, with a named fund and a loan balance. Financial information under the Privacy Act and Standard 8. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A household’s complete 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 redundancy and an insurance exclusion

Employment information not yet public, and a health-related insurance decision. The second is sensitive information. Patient data guide.

Frontier Models (ChatGPT, Claude)	A redundancy and a health exclusion 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 year’s recommendations

The advice, including a referral to a named law practice. The practice’s work product.

Frontier Models (ChatGPT, Claude)	The year’s advice 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

Fees and commission

What the client pays and what the practice earned from the insurer. Disclosure the client is owed, and commercial information about the practice.

Frontier Models (ChatGPT, Claude)	Fee and commission details 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 private enclosure for one spouse

That one client receives something the other does not. A confidence, flagged in writing. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A private arrangement between spouses 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 consent deadline

Operational, and tied to the ongoing fee arrangement the Corporations Act regulates.

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 adviser

Ties the letter to a named adviser bound by the Code of Ethics.

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 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 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 financial advice practices. See how this works in practice: Client review letters on Ermos.

Common questions

Do ChatGPT and Claude use the information in an annual review letter?

Yes, a pasted letter is sent to the AI company's servers and processed there. 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 review letter pairs a home address with fund balances, an insurance exclusion and the commission the practice earned.

Can we use ChatGPT or Claude for an annual review letter if we remove the names?

No, because a review letter is built from details that identify a household even without names. 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. The home address, both partners' exact super balances, the loan balance and a thyroid exclusion can all be matched to the same couple. 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.

Can ChatGPT or Claude remember the details in an annual review letter later?

They can. Both products also have memory features that can carry details from one conversation into later ones, and on a personal account each user decides whether they are switched on (ChatGPT, Claude). An annual letter goes back to the same household every year, so the same details get pasted in again. A client's balances, an insurance exclusion and an expected redundancy could travel into later conversations, with no way for the practice to show the client where they now sit.

How can an advice practice use AI for an annual review letter 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 letter is drafted from the portfolio report, the review file note, last year's letter and the practice's house style. The adviser edits it and signs it before it goes out.

It is a friendly letter. Why does it matter?

Because warmth is the reason it goes into a chatbot, and the content is a year of the client's finances, health and family. The routine letters carry as much as the formal advice, and they are the ones nobody thinks twice about.

Does the fee consent change anything?

The ongoing fee consent requirements in the Corporations Act make the letter part of a regulated arrangement, which is one more reason the record should be the practice's. On Ermos the letter is drafted from the review notes, the portfolio reports and the practice's template, inside the office.

How is a review letter produced on Ermos?

From the file: the portfolio report, the review file note, last year's letter and the practice's house style. The adviser edits and sends it. The client's position never leaves the practice to write 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 financial advice practices AI for financial advisers The Privacy Act, the APPs and AI All professions