Candidate chat messages and AI privacy: what a recruiter text thread gives away in ChatGPT or Claude.

Recruiters work on their phones, and the real conversation with a candidate happens in short messages: a confidence about health or family, a pay figure, a worry about a counter-offer, the client’s secret ceiling shared with a wink. Consultants copy a thread into ChatGPT or Claude to summarise it for the file, to draft the next reply or to check what was promised. This page marks every item in a routine candidate thread that the Privacy Act and anti-discrimination law touch, and shows how the same thread is summarised on Ermos inside the agency.

The candidate chat messages

Kept on the consultant’s phone or a business messaging account through the search, and copied into the candidate file or into an AI tool when someone needs a summary.

Chat with
Tamsin Hartigan · Larkspur Search Partners · 26 to 29 October 2026

Mon 26 Oct, 4.12pm Nina: Can you talk tonight? Something I need to ask before I go further with Northgate.

Mon 26 Oct, 7.48pm Nina: Thanks for the call. To put it in writing:

Mon 26 Oct, 7.52pm Tamsin: Congratulations! Nothing needs to be decided tonight. I will talk to you before anything goes to Marcus.

Tue 27 Oct, 8.05am Nina:

Tue 27 Oct, 8.09am Nina:

Tue 27 Oct, 5.40pm Tamsin:

Wed 28 Oct, 6.30pm Tamsin:

Thu 29 Oct, 7.10am Nina:

An invented chat thread. All people and companies are fictional and the number is a placeholder. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the agency

  1. 1
    The candidate and her mobile number

    A named person and a direct number. Personal information, and the thread’s link to every item below. Privacy Act guide.

  2. 2
    A pregnancy, told in confidence

    Eight weeks, a due date and the fact that her employer does not know. Information about a pregnancy is information about a person’s health, which is likely to be sensitive information under the Privacy Act, and pregnancy is among the attributes the Fair Work Act protects when a prospective employer refuses to employ or offers different terms. The Fair Work Ombudsman says employees and prospective employees are covered by these protections. Patient data guide.

  3. 3
    A job search the employer may suspect

    That the candidate’s manager is already wondering. The candidate’s most exposed fact, and one a leak could make real.

  4. 4
    A target salary and a competing offer

    What the candidate would accept and what a rival has offered. Personal information, and the information that decides a negotiation. Privacy Act guide.

  5. 5
    The client’s hidden ceiling

    A budget the client asked the agency to keep from candidates, shared with one of them. Confidential to the client, and a breach of the instruction on the job order.

  6. 6
    Another candidate named

    That a named person is in the final two. Personal information about a second candidate, told to the first. Privacy Act guide.

  7. 7
    A candid remark about the client’s manager

    The consultant’s private view of a named person, with a request not to repeat it.

  8. 8
    A parent’s hospital visit

    A named hospital, a time and a family member’s follow-up. Health information about a third person, and the candidate’s whereabouts. Patient data guide.

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 agency 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 candidate 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 recruitment agencies.

Common questions

Do ChatGPT and Claude use the information in candidate chat messages?
Yes, everything pasted or uploaded is sent to the AI company's servers and processed there to produce a reply. 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 thread can hold a pregnancy told in confidence, a pay target against a rival offer and a parent's hospital appointment.
Is it safe to paste candidate chat messages into ChatGPT or Claude to summarise them?
No, because the candidate told the consultant things in confidence. 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. In this thread Nina shares an early pregnancy her employer does not know about, and the consultant passes on the client's hidden budget and names a second candidate. Both would reach the AI company.
What if staff use their own ChatGPT or Claude account for candidate chat messages?
The agency cannot see or manage what a consultant sent from a personal account. 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. Consultants work from their phones, so a copied thread with a candidate's pregnancy and a named second candidate could leave without any record at the agency.
How can a recruitment agency use AI for candidate chat messages 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. Ermos summarises the thread for the candidate file using the agency's own file note template. The consultant checks the summary against the messages and signs it before it joins the file.
Is a pregnancy mentioned to a recruiter protected?
The Fair Work Act lists pregnancy among the attributes a prospective employer must not use to refuse to employ someone or to offer different terms, and the Fair Work Ombudsman says employees and prospective employees are covered by these protections. The Privacy Act treats information about health as sensitive information, so the thread needs careful handling in either case.
Are informal messages covered by the Privacy Act?
Personal information is personal information whatever the channel. APP 11.1 requires an APP entity to take reasonable steps to protect it from misuse, interference and loss and from unauthorised access, modification or disclosure, and that includes messages on a phone or a messaging account the agency does not control.
What does the OAIC say about copying a thread into a chatbot?
The OAIC's guidance on commercially available AI products recommends as best practice that organisations do not enter personal information, particularly sensitive information, into publicly available generative AI tools, citing significant and complex privacy risks. A candidate thread with health detail is the kind of content that recommendation addresses.

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 recruitment agencies Resources The Privacy Act, the APPs and AI All professions