Policy comparisons and AI confidentiality: what a quote comparison gives away in ChatGPT or Claude.

Comparing insurer terms is the core of broking and the task most suited to AI, which is why comparison tables are among the first things staff paste into ChatGPT or Claude. A comparison carries the insurers approached, what each quoted, what each excluded, what each pays the brokerage, and the risk information the quotes were based on. This page marks every item in a routine comparison that the Code of Practice and the Privacy Act bear on, and shows how the same comparison is built on Ermos against the brokerage’s own policy library, inside the office.

The policy comparison

Prepared after quotes are obtained, to support the recommendation to the client and the remuneration information the brokerage gives the client.

Quote comparison
, practice package renewal · · 20 October 2026

Risk summary.

Meridian GeneralSouthern Cross InsuranceAnchor Underwriting
Premium
Cyber
Equipment$610,000$610,000$610,000, scanner on schedule
Commission

Notes.

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

What leaves the brokerage

  1. 1
    The client

    A named practice and its company. Everything below is identifiable. Privacy Act guide.

  2. 2
    The risk summary, with a cyber claim

    Turnover, premises, equipment and a prior cyber incident. Commercial information, and a note that the client was successfully phished once, which is the fact a second attacker would most like to know.

  3. 3
    Three insurers’ quotes

    Each insurer’s pricing for this risk. Confidential commercial terms belonging to three insurers as well as the client.

  4. 4
    The exclusions

    What each policy does not cover, matched to the client’s weakest point. Advice in the making, and a map of the client’s uninsured exposure.

  5. 5
    Commissions

    What each insurer pays the brokerage. Remuneration the brokerage sets out in its terms of engagement under section 4.2 of the Code of Practice and, for retail clients, discloses under section 6.1; commercially sensitive to the brokerage and the insurers.

  6. 6
    The broker’s reasoning, including the conflict

    That the recommended insurer also pays the most, and how the broker intends to handle that. Section 5.3 of the Code of Practice on conflicts, worked through in writing.

  7. 7
    A possible sale

    That the practice may be sold next year, and why. Confidential business information about the client and its landlord. Privacy Act guide.

  8. 8
    The broker

    Ties the comparison to a named account manager.

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 insurance brokerages. See how this works in practice: Policy comparison on Ermos.

Common questions

Do ChatGPT and Claude use the information in a policy comparison?
Both receive every line of the comparison 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 policy comparison puts three insurers' quotes, their exclusions and their commissions in one place.
Is it safe to paste a policy comparison into ChatGPT or Claude?
No, because the comparison describes one identifiable practice and what it is exposed to. The risk summary gives the practice's chairs, turnover, leased premises and a prior phishing loss, and the notes record that it may be sold in 2027. Each exclusion is matched to the client's weakest point, so the comparison is advice in the making. 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 be trusted to draft a policy comparison?
Only as a first draft. Language models write fluent text that can contain details that are not in the source, wrong figures and wrong dates. They can also cite a clause that looks real but is not. In a comparison, the costliest errors are a wrong cyber limit, a wrong social engineering sub-limit, an exclusion the wording does not contain and a wrong commission rate. The recommendation rests on the limits and exclusions, and the remuneration disclosure rests on the commission rates. A qualified person should check every row against the quotes, the policy wordings and the client file before the comparison supports any advice.
How can a brokerage use AI for a policy comparison 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 comparison is built from the brokerage's policy library and the client file, with each clause cited to the wording it came from. The broker checks the table and signs off the recommendation.
Comparing policies is exactly what AI is good at. Is there a safe way to do it?
Yes, where the policy wordings and the client's risk information already live. On an Ermos unit the brokerage's policy library and the client file are indexed in the office, and the comparison is built from them, each clause cited to the wording it came from. On Edge the library stays on your own instance in Australia.
Are insurers' quotes confidential?
Quotes are provided to the broker for the client's placement, and insurers treat their pricing and terms as commercially sensitive. A comparison on a frontier model provider's servers holds three insurers' pricing for one risk, which none of them provided for that purpose.
Why does the commission note matter?
Section 4.2 of the Code of Practice requires the brokerage's terms of engagement to set out the remuneration it will earn, and section 6.1 requires the dollar amount of commission to be disclosed where the client is a retail client. Section 5.3 requires conflicts of interest to be identified, disclosed to the client and managed. The note shows the broker working through the conflict in writing. It also shows the brokerage's margins, which are its own business.

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