Most popular AI tools send your data overseas — usually to data centres in the United States — where it is processed and, on consumer tiers, may be retained and used to improve the model. For an Australian accounting, legal, real estate or conveyancing firm, that matters because the moment client information enters an AI system, your obligations under the Privacy Act and the Australian Privacy Principles come with it. Before you adopt any AI tool, you need to know where it hosts data, who can access it, and whether that counts as a cross-border disclosure. This article gives you a checklist to find out.
This is for firm owners and practice managers who want to vet a tool properly before it touches a client file — not become privacy lawyers. It is general guidance, not legal advice. For the underlying obligations, read our companion piece on whether AI is safe for client data under the Australian Privacy Act.
Why data residency matters for Australian professional services
Data sovereignty is the idea that information is subject to the laws of the country it is physically stored and processed in. When your clients' names, financial details, settlement figures or matter notes sit on a server in the US or Europe, that data can fall under foreign laws and foreign government access powers — regardless of your intentions.
Under Australian Privacy Principle 8, sending personal information overseas is a cross-border disclosure. You remain accountable for how that overseas recipient handles it. For professional services firms that also carry confidentiality duties — legal professional privilege, the tax agent code, agency trust obligations — the stakes are higher than for a general business.
The OAIC's general guidance on commercially available AI products is the reference point here: don't enter personal information into publicly available chatbots, and understand how any product handles the data you give it. That guidance is the lens for everything below.
Does ChatGPT store data overseas?
Yes. OpenAI, which runs ChatGPT, is a US company and processes data primarily in the United States. What differs sharply is which tier you use.
- Free and Plus (consumer) tiers: your conversations may be retained and, by default, used to train future models. This is the version you should never put client information into.
- Team, Enterprise and the API: governed by a commercial agreement, data is not used for training by default, and retention is controlled. This is the version a firm can consider — with the right contract and configuration.
The same split applies to most major tools. Anthropic's Claude, Google's Gemini and Microsoft's Copilot all process data overseas on consumer tiers but offer business and enterprise arrangements with contractual data protections. The product name tells you almost nothing; the tier and the contract tell you everything.
How to check where an AI tool stores data
You can establish most of what you need in under an hour, before you ever sign up. Work through these in order.
- Read the Trust or Security Centre page. Reputable vendors publish a dedicated page covering data location, encryption, retention and certifications. If a tool has no such page, treat that as a warning.
- Find the sub-processor list. Most enterprise AI vendors publish a list of the third parties they share data with (cloud hosts, analytics providers). This tells you the real chain of custody, not just the headline vendor.
- Check for an Australian data region. Some platforms let you choose where data is hosted. Microsoft 365 and Google Workspace, for instance, offer Australian data residency options for core services. AI features layered on top may or may not honour that — confirm specifically.
- Look up compliance certifications. ISO 27001 and SOC 2 Type II are the common baselines. They don't tell you where data lives, but they signal a vendor that takes security seriously.
- Read the data processing agreement (DPA), not just the privacy policy. The DPA is the binding document for business use. It should state retention periods, training use, sub-processors and the mechanism for cross-border transfers.
If the answers aren't published, ask directly. A vendor that won't give clear written answers about data location is telling you something.
AI vendor questions about data storage
Send these to any vendor before adoption, and keep the written replies on file. They map directly to your Privacy Act obligations.
| Question | Why it matters |
|---|---|
| In which countries is our data stored and processed? | Determines whether APP 8 cross-border disclosure applies. |
| Is our data used to train your models, now or in future? | Training use means your data leaves your control permanently. |
| How long is data retained, and can we set it to zero retention? | Shorter retention reduces breach exposure. |
| Who are your sub-processors, and where are they located? | The real data chain is longer than the headline vendor. |
| Do you offer an Australian data region? | Keeps data under Australian jurisdiction where available. |
| Will you sign a data processing agreement? | Without it, you have no enforceable contract governing the data. |
| What certifications do you hold (ISO 27001, SOC 2)? | Baseline evidence of a mature security program. |
| What is your breach notification process and timeframe? | You need to meet your own Notifiable Data Breaches obligations. |
A practical data-residency checklist before you adopt a tool
Run every candidate tool through this before it touches a real file:
- Enterprise, Team or API tier only — never a free consumer login for client work.
- Written confirmation that your data is not used for training.
- A signed DPA on file.
- Data location identified and documented, including whether it is overseas.
- If overseas, the transfer treated and recorded as an APP 8 cross-border disclosure.
- Only the fields the task actually needs are sent — redact names and identifiers where the AI doesn't need them to do the job.
- A written data-flow record: what goes in, where it goes, who can see it, how long it is kept.
This is the standard we apply to every build. Prefer tools your firm already licenses and trusts: Microsoft 365 Copilot operating inside your tenant, or Gemini inside Google Workspace, keep data within an environment you already govern rather than introducing a new overseas processor. When you're weighing options, our guide on how to choose an AI automation tool for your firm covers the wider selection criteria.
Australian data residency isn't the whole answer
Here's the honest limit. Choosing a tool with Australian hosting reduces your cross-border exposure, but it does not, on its own, make a tool safe or compliant. A locally hosted tool with weak access controls, no DPA, or a habit of retaining everything indefinitely can be worse than a well-governed overseas enterprise product.
Data residency is one input into a risk assessment, not a pass mark. The bigger questions are whether the data needed to leave your systems at all, who can read it, and what happens if the vendor is breached. Sometimes the right answer is that a task simply shouldn't go through a third-party AI — keep it in-house, or don't automate it.
Turning this into firm policy
A one-off check for one tool solves today's problem. What protects you long term is a written position that every staff member follows — which tools are approved, which tiers, what data may and may not be entered, and who signs off on new tools. Without it you get shadow IT: people pasting client details into whatever free chatbot is open in another tab.
If you don't have that document yet, our AI policy template for Australian law firms gives you a practical checklist that adapts easily to accounting, real estate and conveyancing practices. Pair it with the vendor questions above and you have a repeatable process for vetting any new tool.
Knowing where your data goes is the foundation of using AI responsibly — once that's settled, the productivity gains are real. Our free guide, 10 AI Workflows to Save 10+ Hours a Week, shows the everyday automations Australian firms are putting in place, with data handling built in from the start rather than bolted on afterwards.
Common questions
Where does AI send your data if you're an Australian firm?
Most mainstream AI tools process data in overseas data centres, commonly in the United States. On free consumer tiers that data may also be retained and used to train the model. Enterprise, Team and API tiers are governed by contract and usually keep data out of training. Always confirm the exact location and tier in writing before using a tool for client work.
Does ChatGPT store data overseas?
Yes. OpenAI processes ChatGPT data primarily in the United States. On the free and Plus tiers, conversations may be retained and used to improve the model by default. Team, Enterprise and API arrangements are governed by a commercial agreement, exclude training use by default, and give you control over retention. Never enter client information into the consumer tiers.
Is Australian data residency enough to make an AI tool compliant?
No. Australian hosting reduces cross-border disclosure risk under Australian Privacy Principle 8, but it's only one factor. A locally hosted tool can still have weak access controls, indefinite retention or no data processing agreement. Residency is an input into a proper risk assessment, not a compliance pass. Assess access, retention, contracts and whether the data needed to leave your systems at all.
What should I ask an AI vendor about data storage?
Ask where data is stored and processed, whether it's used to train models, how long it's retained, who the sub-processors are, whether an Australian region is available, and whether they'll sign a data processing agreement. Also ask about certifications like ISO 27001 and SOC 2, and their breach notification process. Keep the written answers on file as part of your records.
How can I find out where an AI tool stores data before signing up?
Start with the vendor's Trust or Security Centre page, which reputable providers publish. Check their sub-processor list for the full data chain, look for an Australian data region option, and read the data processing agreement rather than just the privacy policy. If the information isn't published, email the vendor directly and treat any reluctance to answer as a warning sign.
Want these ideas working in your firm? We build controlled AI workflows for Australian professional services firms — starting with a free automation audit.