Client emails and AI confidentiality: what a reassurance email gives away in ChatGPT or Claude.

When markets fall, advisers write a lot of emails fast, and each one has to be accurate, calm and personal. That is exactly the brief people give ChatGPT or Claude. The email carries the client’s holdings, what they have lost, the advice, and often a reference to how the client behaved last time. This page marks every item in a routine reassurance email that the Code of Ethics and the Privacy Act protect, and shows how the same email is drafted on Ermos from the client file inside the practice.

The client email during a market fall

Sent in response to a client’s call or email during a market downturn, usually the same day, and often to many clients at once with personal figures in each.

To: Lucy Harrington¹
From: Nadia Reyes, Harbourview Wealth Partners⁸
Subject: Re: should we sell?

Dear Lucy,

I understand the worry, and I am glad you emailed before doing anything.

Your consolidated super balance is $184,100 this morning, down from $198,700 at the review, a fall of 7.3% over three weeks. Tom’s is $409,800, down 6.5%. The Tesla shares are $24,900, down a further 21%.²

Our advice is to hold. Your super is invested for a horizon of twenty years or more, the balanced option you chose is doing what it is designed to do in a fall of this size, and selling now would lock in the loss. The Tesla position is the exception: we have recommended selling it since last year, and this fall does not change that recommendation.³

I know March 2020 was difficult and that you sold at the bottom then. You told us at the fact find that you did not want to be in that position again, and the balanced option was chosen partly for that reason. Please call me before making any change.⁴

One practical point: with Tom’s redundancy expected in December, we want the offset account intact. Nothing in the super accounts affects that.⁵

If it would help, I can arrange a call with Tom as well, though I note your earlier request about keeping some matters between us; this one is fine to share if you wish.⁶

Kind regards,
Nadia⁸
This email is based on the circumstances recorded in your file and should be read with your Statement of Advice and Records of Advice. Please tell us if anything has changed.⁷

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

What leaves the practice

The client

Named. Everything below is identifiable. 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 and losses, to the day

Two people’s superannuation and a share holding, with current values. Financial information under the Privacy Act and Standard 8. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	Balances transmitted; on consumer tiers they 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 advice

The practice’s recommendation in a live situation, including a standing recommendation the client has rejected. Advice, documented.

Frontier Models (ChatGPT, Claude)	Live 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

The client’s history with loss

A reference to what she did in 2020 and what she said at the fact find. Personal, and the kind of detail that would embarrass her anywhere else. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A client’s history with loss 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 redundancy

That her husband’s job is ending in December. Confidential employment information, in an email about something else. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A redundancy 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 reference to a private arrangement

An allusion to matters kept between the adviser and one spouse. A confidence, acknowledged in writing. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	A confidence acknowledged 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

The disclaimer

The practice’s standard wording about the status of the email. The sentence the practice relies on, now alongside the figures it qualifies.

Frontier Models (ChatGPT, Claude)	Your protection transmitted with what it protects
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 email 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 correspondence on Ermos.

Common questions

Do ChatGPT and Claude use the information in a client email about a market fall?

Yes, a pasted email 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. It carries balances to the day, the advice to hold and a reference to the client's behaviour in 2020.

Can we use ChatGPT or Claude for a client email if we remove the client's name?

No, because the figures and the history in the email still describe one person. 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. Two super balances to the dollar, a share holding, a reference to selling in March 2020 and a husband's redundancy would let anyone who knows the family name the client. 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.

What if staff use their own ChatGPT or Claude account for a client email?

Then each email leaves the practice's control the moment it is pasted into a tool run by OpenAI or Anthropic. 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. Both companies can also review content that is flagged for safety or policy reasons. The practice has no administrator access to a personal account. In a downturn that could mean dozens of emails, each with balances to the day, the advice to hold and personal history.

How can an advice practice use AI for a client 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's holdings and the file note in the practice's own tone. The adviser personalises it and checks the figures before sending.

Advisers send dozens of these in a downturn. Is there a safe way to use AI for them?

Yes: draft them where the client data already lives. On an Ermos unit the email is drafted from the client's holdings and file note, in the practice's tone, and the adviser personalises and sends it. Dozens of emails with personal figures never leave the practice to be written.

Is the reference to 2020 really a problem?

It is personal information about how the client handles money and stress, recorded under Standard 8 because it matters to the advice. Inside the practice it is good advice. On a frontier model provider's servers it is a character sketch of a named person.

Does the disclaimer help?

It tells the client how to read the email alongside the formal advice. It does nothing about where the email was drafted.

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