Visa application forms and AI data leakage: what a family data form gives away in ChatGPT or Claude.

Before an agent lodges a visa application, the client completes a data form for the whole family: who they are, where they have lived, what they do for work, their health and any convictions. It is the densest document a migration practice holds, and the one staff most often paste into ChatGPT or Claude to extract fields into an application or to spot gaps. This page marks every item in a typical family data form that the Code of Conduct and the Privacy Act bear on, and shows how the same extraction happens on Ermos inside the practice.

The visa application data form

Completed by the client after the service agreement is signed and before the application is prepared. Section 36 of the Code requires the agent to verify the client’s identity, and the form feeds the application the agent prepares.

Family data form
Kingfisher Lane Migration · Confidential · received 12 October 2026

Main applicant
Partner
Children
Current visa
Employment
Health
Character
Visa history
Police certificates

An invented data form. The passport numbers are placeholders and the family, employer and doctor do not exist. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the practice

  1. 1
    The main applicant’s identity

    Full name, date of birth, nationality and passport number. The identifiers an identity thief needs, and the details section 36 of the Code requires the agent to verify before giving immigration assistance. Privacy Act guide.

  2. 2
    The partner’s identity

    A second adult’s name, date of birth, nationality and passport number. Personal information about a person who may never have dealt with the practice directly. Privacy Act guide.

  3. 3
    The children

    Two named minors, their birth dates and the town they go to school in. Children’s personal information, and where they are on weekdays. Privacy Act guide.

  4. 4
    Visa status and expiry

    The visa the family holds, when it ends and what the family hopes to apply for. The deadline that drives the whole file. Privacy Act guide.

  5. 5
    Employer, location and salary

    A named employer, how long she has worked there and what she is paid. Employment and financial information the application will rely on. Privacy Act guide.

  6. 6
    A child’s health condition

    A diagnosis, a medication and a named doctor. Health information is sensitive information where the Privacy Act applies, and health matters also feed the health requirement for the visa. Patient data guide.

  7. 7
    A criminal conviction

    A named person’s conviction, the place, the date and the penalty. A criminal record is sensitive information under the Privacy Act, and the character answers must be accurate because the Code prohibits false or misleading statements to government officials. Privacy Act guide.

  8. 8
    A third party’s visa refusal

    A relative’s name and a refused visa. Personal information about someone who is not the client and has not agreed to its use. Privacy Act guide.

  9. 9
    The police certificate countries

    Which countries the adults have lived in. The rule for when Home Affairs asks for police certificates is public, but the countries disclose where this family has lived and can point to national and ethnic background. Privacy Act 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 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 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 migration practices.

Common questions

Do ChatGPT and Claude use the information in a visa application data form?
Yes, anything pasted from the form 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. One form carries two passport numbers, two children's names and birth dates, a child's asthma and a drink-driving conviction.
Can we use ChatGPT or Claude for a family data form if we remove the names?
No, because the details still describe one family. 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 adults' birth dates, a regional town where both children go to school, a named employer and a salary identify them without a name, and the passport numbers need no inference at all. 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 deleting the chat remove a visa application data form from ChatGPT or Claude?
Not reliably. Deleting a chat removes it from the user's history, but OpenAI and Anthropic each say they can keep a deleted chat on their systems for up to 30 days, and longer where safety or legal obligations apply (OpenAI, Anthropic). Both products also have memory features that can carry details from one conversation into later ones, and on a personal account each user decides whether they are switched on (ChatGPT, Claude). A delete button gives the practice no way to show the family that their passport numbers, the children's school or the conviction are gone.
How can a migration practice use AI for a visa application data form 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 application fields are drafted from the client file and the practice's own records. A registered migration agent checks every field against the original documents before lodging.
Why is a family data form so risky to paste into an AI tool?
Because it collects everything in one place: identity documents for each adult, the children, visa status, employer and salary, health answers and character declarations. Most documents hold one kind of sensitive information. This one holds the set an identity thief, a scammer or a hostile relative would want, for several people, only one of whom signed the practice's consent.
Does a conviction on the form change what the practice must do?
The Code of Conduct prohibits making false or misleading statements to government officials, and requires reasonable steps to check statements the agent is unsure of. A conviction therefore has to be declared accurately. Separately, a criminal record is sensitive information under the Privacy Act, which applies only to practices outside the small business exemption or caught by an exception.
Staff paste the form to fill in the application. Is the purpose the problem?
The purpose is sensible; the destination is the problem. On an Ermos unit the form is read on hardware in the office and the fields go into the practice's systems. On Edge the form stays on the practice's own instance in Australia. Either way the agent checks every field against the original documents before lodging.

See Dominion answering from your own documents.

A demonstration takes 30 minutes, uses no client data, and comes with no obligation.

Book a demo