Text and chat messages and AI data leakage: what a pasted message thread gives away in ChatGPT or Claude.

Texts, SMS and chat threads with clients and colleagues are informal, fast and full of detail, and staff paste them into AI tools to summarise a conversation or draft a reply. Threads carry first names, plans, addresses and photos of other documents. These pages mark up real message threads across professions.

Aged care

The staff group chat handover

Three days of handover: an allegation against a colleague, injury photos, a dying resident, a shared login and an open police matter.

See what leaves →

Healthcare

The appointment and recall SMS thread

A name and number, a doctor, a blood test, a family cancer history and a request to stop texting. A recall, in eight lines.

See what leaves →

Dental practices

The appointment texts and web chat

Appointment times, a spouse explaining a medication, a pregnancy, a card payment and a patient who is furious about a fee.

See what leaves →

Conveyancers

The buyer and seller chat messages

An alarm code, a bereavement, a pregnancy and a late request to send the proceeds to someone else’s account.

See what leaves →

Migration agents

The WhatsApp thread with a client

A passport photo, a bank transfer receipt, a wife on chemotherapy and a job search the client has not told his employer about. All typed between meetings.

See what leaves →

Real estate agents

The landlord and tenant text messages

An owner who wants a tenant listed, a tenant in difficulty, a safety disclosure and a staff mistake, in one thread.

See what leaves →

Recruiters

The candidate chat messages

A pregnancy, a job search kept secret, a pay figure, the client’s budget and a father’s hospital visit. All in a text thread.

See what leaves →

Common questions

Is it safe to put a message thread into ChatGPT or Claude?
Not when it holds client, patient or participant information. Chat and text threads are informal, so they carry first names, plans, addresses and photos of other documents. 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. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product.
Do ChatGPT and Claude use the information in a message thread?
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. Whatever the plan, the AI company receives everything pasted or uploaded to produce its answer.
Can I remove the names from a message thread before using ChatGPT or Claude?
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. Chat and text threads are informal, so they carry first names, plans, addresses and photos of other documents.
How can a practice use AI on message threads without sending them 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. A person still checks the result against the source before it is relied on.
What do the pages about message threads show?
Each page takes one kind of message thread from one profession, invented in every detail, and marks every item the business is obliged to protect. Hover or tap an item for the obligation it touches and a three-way comparison of Frontier Models (ChatGPT, Claude), Ermos Edge and Ermos Dominion. Items are rated honestly, so some are neutral or not applicable and some are fine to share.
Are the example message threads real?
No. Every document is invented, including the names, figures and references. Identifiers use published example values or obvious placeholders, so nothing can match a real person or business.
Which professions are covered?
Lawyers, accountants, financial advisers, insurance and mortgage brokers, healthcare and dental practices, aged care and NDIS services, conveyancers, migration agents, real estate agencies and recruiters. Each profession's page lists all of its documents.

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 by profession The Privacy Act, the APPs and AI Compliance guides