Private ChatGPT for business: the Australian options compared

A private ChatGPT for business is an AI assistant your organisation controls: staff sign in to a managed workspace, the vendor doesn't train its models on what they type, and you know where the data is stored and processed. Several products meet the first two tests. Far fewer meet the third if you need the data to stay in Australia.

What a private ChatGPT for business usually means#

When an Australian business asks for a private ChatGPT, it rarely means OpenAI's product specifically. It means a ChatGPT-style assistant that passes four tests:

1. No training on your data. Prompts, files and responses aren't used to improve the vendor's models, by default and in the contract.
2. Admin control. The organisation decides who has access, can remove leavers, and isn't relying on staff using personal accounts.
3. Known data location. You can say where conversations are stored and where the AI processing (inference) runs. These are often different places.
4. Retention you understand. You know how long prompts and files are kept, and who can delete them.

Personal accounts usually fail these tests. OpenAI's pricing page says content from its individual plans can be used to train its models unless the user opts out, and Anthropic's pricing page lists model training on its individual Claude plans as opt-out too. The Office of the Australian Information Commissioner (OAIC) recommends that organisations don't enter personal information, particularly sensitive information, into publicly available generative AI tools.

Your options in Australia#

ChatGPT Business and Enterprise (OpenAI). OpenAI states that it doesn't train on business data by default. Business is self-serve for two or more users, with SAML SSO, and workspace admins can control how long data is retained. Storage at rest in Australia is offered for eligible new Enterprise and Edu workspaces; OpenAI doesn't list Business as eligible. OpenAI's inference residency, which keeps model processing in-region, currently covers Europe, the US and the UAE, so ChatGPT prompts aren't processed in Australia even when they're stored here.

Microsoft 365 Copilot (being renamed Microsoft Copilot). A paid add-on to eligible Microsoft 365 plans that works inside apps such as Word and Teams and draws on your emails, chats and documents through Microsoft Graph. Microsoft states that prompts, responses and data accessed through Microsoft Graph aren't used to train foundation models, and interaction data is stored with your other Microsoft 365 content under its data residency commitments. Outside the EU, Microsoft says queries may be processed in the US, EU or other regions. In April 2026 it said local processing in Australia is expected by the end of 2026; its announcement describes the covered data as prompts and responses from Copilot models hosted on Azure OpenAI. Anthropic's Claude models, which Microsoft turns on by default for most commercial customers outside the EU, EFTA and UK, are currently excluded from in-country processing commitments.

Claude Team and Enterprise (Anthropic). Anthropic doesn't train on content from its commercial plans by default, although chats you submit as feedback may be used. Team includes SSO; Enterprise adds audit logs, SCIM and custom retention controls. Anthropic says data for its commercial products, such as Claude Enterprise and the API, is stored in the US, and traffic may be routed to countries in the US, Europe, Asia and Australia. Anthropic lists its regional data residency options through cloud platforms such as Amazon Bedrock and Google Cloud Vertex AI.

AusGPT. Claude models through Amazon Bedrock's Australian cross-region inference profile, which routes requests only between AWS Sydney and Melbourne. Conversations and documents are stored on AWS in Sydney and never used to train AI models. AusGPT is a separate product and is not affiliated with OpenAI.

How they compare on privacy#

Vendor details as of October 2026, from each vendor's published documentation (see sources below). These change often, so check before you decide.

Why AusGPT fits teams that need data in Australia#

AusGPT's position is simple: Claude, processed in Australia.

• Processing and storage in one place. Every AI request goes through Bedrock's Australian profile, so prompts and documents are processed in Sydney and Melbourne and stored in Sydney. Dictation uses Amazon Transcribe in Sydney.
• No training, no sharing with the model maker. Amazon Bedrock states that customer inputs and outputs aren't shared with model providers or used to train models, and model providers, including Anthropic, have no access to customer prompts on Bedrock.
• Ready for a team on day one. Team workspaces with member invitations, conversation sharing, a shared prompt library with folders, and document chat for PDF, Word and image files.
• Predictable cost. A$29 per user per month on Business, with 2,000 messages per user each month. Enterprise adds SSO, audit logs, white labelling and custom integrations with your business systems.

It isn't the right fit for everyone. If your team wants AI inside Outlook and Word, drawing on SharePoint and email, Microsoft 365 Copilot does things AusGPT doesn't. AWS is a US-headquartered company, so onshore processing keeps data in Australian data centres but doesn't change who operates them. And a brand-new Claude model may reach AusGPT later than Anthropic's own apps, because we only offer models available on the Australian profile.

Questions to ask any vendor#

Use these whichever product you choose. For a fuller checklist, see our guide to secure AI for business.

• Is our data used to train models? Is "no training" in the contract, or only a default setting?
• Where is data stored at rest, and where does inference run? Are both in Australia?
• Which features fall outside those commitments? OpenAI, for example, excludes data sent through external integrations and web search from its residency scope.
• Which subprocessors or model providers can see prompts and files?
• How long are conversations and files kept, and can admins delete them?
• Can we enforce sign-in, remove leavers and see who has access?

Under Australian Privacy Principle 8, an organisation that discloses personal information to an overseas recipient generally remains accountable for how that recipient handles it. Keeping storage and processing onshore won't settle every privacy question, but it makes that one much easier to answer. See how AusGPT keeps data in Australia or contact us for the detail your security review needs.