Advice letters and AI confidentiality: what privileged advice gives away in ChatGPT or Claude.

The letter of advice is the document privilege exists to protect, and the one most often pasted into ChatGPT or Claude to tighten the prose. It carries everything: who the client is, what happened, who the other side is, how strong the case is and what the firm proposes to do about it. This page marks every item in a routine employment advice that rule 9 and legal professional privilege protect, explains the AI confidentiality question each one raises, and shows how the same letter is drafted on Ermos inside the firm.

The letter of advice

Sent once the solicitor has reviewed the file and formed a view, and relied on by the client in deciding what to do.

Carrick Doyle Lawyers
Privileged and confidential

Ms Marguerite Soto¹
12 Bayview Terrace, Coogee NSW 2034¹

Re: Termination of employment by Greystone Property Group Pty Ltd²

Dear Marguerite,

You asked for our advice on whether you have a claim arising from your dismissal on 4 September 2026³. We set out the facts as you have told them to us, our view of your prospects, and what we recommend.

The facts. You were employed as Finance Manager from March 2019 on a salary of $168,000. In June you raised concerns with the CFO, Alan Whitcombe, about the capitalisation of fit-out costs in the half-year accounts. On 4 September you were told your role was redundant. No other role in the finance team was made redundant, and a Financial Controller position was advertised on 18 September.³

You have told us that in August you were diagnosed with an anxiety disorder and took eleven days of personal leave, which Mr Whitcombe described in a meeting as “convenient”.⁴

Our view. In our opinion you have reasonable prospects in a general protections claim under the Fair Work Act, on the basis that the dismissal was taken because you exercised a workplace right in raising the accounting concern. The timing, the absence of other redundancies and the subsequent advertisement support an inference the employer will have to rebut. We assess the prospects of establishing the claim at better than even, and a discrimination claim in relation to your leave as weaker but worth pleading in the alternative.⁵

Recommendation. Lodge the general protections application before the 21-day limit expires on 25 September. Before conciliation, we would open at twelve months’ salary and accept nine. We do not recommend reinstatement as a remedy, for reasons we can discuss.⁶

Our fees to conciliation are estimated at $14,000 plus GST. If the matter proceeds to hearing, we will provide a further estimate.⁷

Eleanor Vance⁸
Partner⁸

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

What leaves the firm

The client and her home

Named, at a residential address, in an employment dispute. Personal information under the Privacy Act and confidential under rule 9. Rule 9 confidentiality 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 opponent

The employer, named, and the fact it is facing a claim. Confidential to the client, and sensitive to a company that has not been served yet. Rule 9 confidentiality guide.

Frontier Models (ChatGPT, Claude)	A company’s exposure transmitted before it has been served
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 account of the facts

Salary, the accounting concern she raised, the CFO’s name and the sequence of events. Communications made for the dominant purpose of legal advice, which is what privilege attaches to. Legal privilege guide.

Frontier Models (ChatGPT, Claude)	Privileged communications 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

A health condition

A diagnosis, leave taken and a manager’s remark about it. Sensitive information under the Privacy Act, and the most personal paragraph in the letter. Privacy Act guide.

Frontier Models (ChatGPT, Claude)	Sensitive health information 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 firm’s assessment of prospects

The firm’s opinion on the claim’s strength and its weakest element. The core of privileged advice, and the paragraph the other side would most like to read. Legal privilege guide.

Frontier Models (ChatGPT, Claude)	Your assessment of the case 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

Strategy and settlement range

Open at twelve, accept nine, no reinstatement. The client’s negotiating position, written down before negotiation has begun. Legal privilege guide.

Frontier Models (ChatGPT, Claude)	The settlement range transmitted before negotiation begins
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

The firm’s pricing for the matter, on a third party’s servers with the client it was quoted to.

Frontier Models (ChatGPT, Claude)	Your pricing transmitted with the client it was quoted to
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 partner

Ties the advice to a solicitor bound by rule 9, and makes any leak traceable to the practice. Law Society AI guidance.

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 firm 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 matter 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 law firms. See how this works in practice: Legal research on Ermos.

Common questions

Do ChatGPT and Claude use the information in a letter of advice?

Yes, whatever is pasted in is sent to the AI company's servers to be processed. 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 letter of advice carries the assessment of prospects, the settlement range and the client's diagnosis.

Can we use ChatGPT or Claude for a letter of advice if we remove the names?

Taking out the names does not make a letter of advice safe, because the facts still point to 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. The dismissal date, the salary, the job title and the accounting concern identify the client, and the diagnosis is sensitive under any name. 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 letter of advice?

Not on its own. The firm would still be sending the letter to a third-party service and relying on that service's terms. 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. In a letter of advice that content is the firm's view of the client's prospects and her settlement range, which is what privilege exists to protect. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Our comparisons with ChatGPT Enterprise and Claude weigh the trade-offs.

How can a law firm use AI for a letter of advice 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 solicitor asks for the draft in plain language, and it comes from the matter file and the firm's own precedent letters. The solicitor checks every statement before signing the letter.

Does pasting an advice letter into ChatGPT or Claude waive privilege?

Privilege turns on confidentiality, so the careful question is whether disclosure to a third-party service is consistent with maintaining it. That analysis is contested for cloud tools, and the Law Society guidance warns that placing client information in a publicly available system is akin to putting it in the public domain, with privilege among the things a client may lose. For a unit inside the firm the question does not arise, because no third party receives anything.

Can AI help draft advice without the facts leaving the firm?

On Ermos, yes. The unit works from the firm's own precedents and the matter file on your network, and every answer carries the excerpts it drew on. The research and the drafting happen inside the firm; the solicitor's judgement and the verification remain the solicitor's.

What about the employer's name?

The employer is not the firm's client and the firm owes it no duty of confidence. The client does: an advice letter naming the opponent and the claim, on a third party's servers before service, is a disclosure the client did not authorise.

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 law firms AI for law firms The Law Society AI guidance All professions