Inspection reports and AI privacy: what a rental condition report gives away in ChatGPT or Claude.

After a routine inspection the property manager has rough notes, a phone full of photos and an owner waiting for a report. Pasting the notes into ChatGPT or Claude to tidy them into a clear report looks like a harmless way to save an hour. The report describes the inside of someone’s home: who lives there, the key safe and alarm codes, what is on the kitchen bench, how the tenant seemed and what the owner will be asked to pay for. This page marks every item in a routine inspection report and shows how the same report is written on Ermos inside the agency.

The routine inspection and condition report

Written after each routine inspection and sent to the owner, with photos, so that repairs can be approved and the condition of the property is on record.

Routine inspection report
4/31 Lorikeet Road, Summer Hill · Tuesday 13 October 2026, 10.15am to 10.40am · Owner: Gerald Penhall

Attendance.

Access and security.

General.

Defects.

Observations.

The tenant.

Photos.

Recommended actions.

An invented inspection report. The people, the address and the key safe code are fictional. Hover or tap any highlighted item for the analysis; the full list is below.

What leaves the agency

  1. 1
    The tenant and her household

    A named tenant and that she lives alone after a separation. Personal information about a tenant, with a detail she gave in her own home. Privacy Act guide.

  2. 2
    The checklist headings and ticks

    Standard inspection wording: smoke alarm, locks, floors, garden. Generic text with no personal data, which an agency could safely share as a template.

  3. 3
    Key safe, alarm code and spare key

    The ways into the home. Pasted into a tool that stores prompts, they sit with the address beside them, and the tenant was told nothing of it. Privacy Act guide.

  4. 4
    Defects in the owner’s property

    Mould, an ageing hot water unit and a loose handrail. Facts about the owner’s asset that bear on repairs and on liability if someone is hurt.

  5. 5
    An unapproved occupant and an unapproved pet

    Another person’s belongings in the hallway and a dog bowl where the lease permits a cat. Observations about who is living in the home that could be wrong and could count against the tenant. Privacy Act guide.

  6. 6
    The property manager’s impressions of the tenant

    A tenant’s sleep, mood and unopened mail, including debt collection. Written for the file and the owner, and personal and financial information about the tenant. Privacy Act guide.

  7. 7
    Photos of private belongings

    Photos of bedrooms, medication and family photos. The inside of a person’s home, with health information visible in at least one frame. Patient data guide.

  8. 8
    Recommended repairs and costs

    What the owner is asked to approve and what it is expected to cost. The owner’s money and the agency’s advice to a client.

  9. 9
    The property manager

    Ties the report to a named employee of the agency.

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 property 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 real estate agencies.

Common questions

Do ChatGPT and Claude use the information in a routine inspection report?
Notes pasted from the report go to the AI company's servers and are 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. This report holds a key safe code beside the address, a note on where the spare key is hidden, black mould in the bathroom, medication on the kitchen bench and letters from a debt collector.
Is it safe to paste a routine inspection report into ChatGPT or Claude?
No, because the report describes the inside of a named tenant's home. 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. It lists the key safe code, the alarm code and the spare key beside the address, and it records the tenant's mood, her unopened mail and photos that show medication. 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.
Does deleting the chat remove a routine inspection report from ChatGPT or Claude?
Not on its own. 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). For this report, the key safe code, the alarm code and the tenant's circumstances could stay on the AI company's systems for that period after the chat leaves the user's history. A report that gives the means of entry to a home is a poor candidate for that.
How can a real estate agency use AI for a routine inspection report 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 report is drafted from the property manager's inspection notes and the agency's own report template. The property manager checks every observation against what was seen at the inspection, then signs it.
Are photos taken at an inspection personal information?
Photos that show a person's belongings, medication, documents or family are personal information about the tenant. NSW Fair Trading's guidance on tenancy applications says agents should explain why information is collected and how it will be used, and the same discipline is sensible for photos kept after an inspection.
Why are key safe and alarm codes a leak risk in a report?
Because a code is a working key. A report that lists the address, the key safe code and where the spare key is hidden gives anyone who reaches it the means to enter. Codes belong in the agency's secured systems, and sending them to a service that stores prompts puts them somewhere the agency does not control.
Does the Privacy Act cover a tenant's information in an inspection report?
It is personal information. Whether the Act binds a particular agency depends on its turnover and activities: the OAIC says most small businesses with turnover of $3 million or less are outside it, except operators of a residential tenancy database and certain others. A small agency that is an AML/CTF reporting entity is covered only for its AML/CTF activities.

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