Australian Privacy Principles and AI Tools: A Plain-English Guide for SMBs

If you've started using AI tools in your business and you're not sure whether you're doing it correctly under Australian privacy law, you're in good company. Most SMB owners using tools like ChatGPT, Microsoft Copilot, or AI-powered customer service platforms haven't had a formal privacy briefing on what that use actually requires. They've just started using the tools, because the tools are useful.

The problem is that Australian privacy law, specifically the Privacy Act 1988 and the Australian Privacy Principles that sit inside it, does apply to how you collect, use, store, and share personal information, including when an AI tool is doing the collecting and sharing on your behalf. Understanding which principles apply to AI use, and what they actually ask of you, is the first step to using these tools with confidence rather than anxiety.

For a broader overview of how the Privacy Act applies to AI use in general, see our Privacy Act and AI guide. This article goes one level deeper, working through the specific Australian Privacy Principles that matter most for AI tool use, one at a time.

What Are the Australian Privacy Principles?

The Australian Privacy Principles are 13 legally binding rules that sit inside the Privacy Act 1988. They govern how organisations collect, hold, use, and disclose personal information about individuals. Personal information means anything that could identify a specific person, including a name, email address, phone number, health record, or financial detail.

Not every business in Australia is covered. If your business has an annual turnover of more than $3 million, you are covered by the Privacy Act regardless of your industry. Some smaller businesses are also covered depending on their activities, including businesses that handle health information, operate as credit providers, or are contracted service providers to government agencies. If you're unsure whether the Act applies to your business, the OAIC guidance on AI is a useful starting point, and a privacy lawyer can confirm your status in one short consultation.

Once the Privacy Act applies to your business, all 13 APPs apply. This guide focuses on the six that are most directly triggered by ordinary AI tool use.

APP 1: Open and Transparent Management of Personal Information

APP 1 requires organisations to manage personal information in an open and transparent way. The OAIC's full APP 1 guidance is at oaic.gov.au. In practice, this means having a Privacy Policy that accurately describes how you handle personal information, and keeping that policy current. It also means that if you ask someone for their personal information, you're not doing anything with it that would surprise them or that they haven't been told about.

Where AI tools come in: if your business is using an AI tool that processes customer data, staff data, or any other personal information, the OAIC's guidance indicates your Privacy Policy should reflect that. A policy written before your business used any AI tools is likely out of date. You should also be aware that the upcoming amendments to the Privacy Act will introduce new transparency requirements specifically for automated decision-making, which includes many AI functions. The December 2026 ADM deadline is the most immediate compliance pressure for businesses using AI in ways that affect individuals.

APP 3: Collection of Solicited Personal Information

APP 3 sets out the rules for what personal information you can collect. The core rule is that you should only collect personal information that is reasonably necessary for your functions or activities. You should not collect more than you need, and you should collect it by lawful and fair means.

AI tools create a specific risk here because they are very good at collecting and processing large volumes of data, often more than you'd consciously choose to collect by hand. An AI customer service chatbot, for example, might log full conversation transcripts including sensitive disclosures that a customer makes incidentally. An AI note-taking tool in a meeting might capture everything said, including information about third parties who weren't expecting to be recorded.

Applying APP 3 to AI tool use means thinking about what the tool is actually collecting, not just what you intended it to collect, and confirming that collection is limited to what your business genuinely needs.

APP 5: Notification of the Collection of Personal Information

Under APP 5, the OAIC's guidance outlines that businesses are expected to take reasonable steps to notify individuals of certain things: who you are, what you'll use the information for, whether you'll disclose it to third parties, and whether you're likely to disclose it overseas. The OAIC's guidance indicates this notification should happen at the time of collection, or as soon as practicable afterwards. See APP 5 guidance at oaic.gov.au.

For AI tools, this is frequently a gap. If your business website has an AI chatbot and a customer types their name and complaint into it, does your website make it clear that information is being collected, who holds it, and where it goes? Many SMBs have deployed AI chat tools without updating their website's privacy notice or collection statement to account for the new data flows.

APP 6: Use or Disclosure of Personal Information

APP 6 sets out when you can use or disclose personal information that you've already collected. The primary rule is that personal information should only be used or disclosed for the primary purpose for which it was collected, or for a secondary purpose that the individual would reasonably expect, or where they've consented to the secondary use.

AI tools regularly create APP 6 tension because they tend to do things with data beyond what a customer originally expected when they handed over their details. For example, a customer who gave you their email address to receive an invoice would not necessarily expect that email to be processed by an AI tool that analyses customer sentiment or categorises support cases. Whether that constitutes a breach of APP 6 depends on the specific circumstances, but the question is worth asking.

If you're also sharing customer data with an AI vendor, it's worth checking whether that vendor uses your data to train their models. Some vendors do, and some don't. This matters under APP 6 because model training is a use of the personal information beyond the immediate transaction, and it's not something most customers would expect. See the customer data and ChatGPT guide for a specific look at how this plays out with one of the most commonly used tools.

APP 8: Cross-Border Disclosure of Personal Information

APP 8 is the Australian Privacy Principle most directly triggered by using cloud-based AI tools, and it's the one Australian SMBs most often overlook. APP 8 requires that before you disclose personal information to someone overseas, you take reasonable steps to ensure that the overseas recipient will handle the information in a way that is at least as protective as the Australian Privacy Principles require.

Almost every major AI platform, including ChatGPT (OpenAI, US), Microsoft Copilot (US), Google Gemini (US), and Anthropic's Claude (US), processes data on servers outside Australia. When you input personal information about a customer, employee, or any individual into these tools, that constitutes a cross-border disclosure under Australian law. APP 8 applies.

There is an exception under APP 8.2(b): if the individual has consented to the overseas disclosure after being informed that APP protections may not apply, the obligation on your business is reduced. This is why some businesses are updating their privacy notices to include a cross-border disclosure consent clause specifically for AI tool use. It's a practical workaround, but it requires that the consent is genuinely informed and freely given, not buried in a terms-of-service click-through.

APP 11: Security of Personal Information

APP 11 requires that you take reasonable steps to protect personal information from misuse, interference, loss, unauthorised access, modification, or disclosure. It also requires you to destroy or de-identify personal information when you no longer need it, and when you are no longer required by law to retain it.

AI tools introduce specific security considerations. If staff are copying and pasting customer data into a public AI chatbot interface, that data is being transmitted to and potentially retained by a third party. If your business uses an AI tool with weak access controls, an employee or contractor could access personal information they shouldn't be able to reach. If the AI vendor experiences a data breach, your customers' data may be at risk even though you did nothing wrong on your own systems.

APP 11 doesn't require perfection, it requires reasonable steps. For most SMBs, reasonable steps include choosing AI vendors with credible security certifications, configuring tools to avoid unnecessary data retention, and training staff on what personal information should and shouldn't be entered into AI tools.

Which Industries Face Extra Scrutiny

Every business covered by the Privacy Act is expected to apply the APPs described above. But some industries face additional obligations or higher-stakes consequences when AI tools are involved.

Healthcare and allied health. Health information is sensitive information under the Privacy Act, and it has a higher collection threshold and stricter use limits. Using AI tools that process patient notes, clinical records, or health-related communications requires careful attention to APPs 3, 6, and 11 in particular. Medicare and private health data stored in overseas AI systems is a specific concern.

Accounting, bookkeeping, and financial services. AI tools used for bookkeeping, financial reporting, or client advisory functions often process detailed financial information. ASIC and the Tax Practitioners Board have their own obligations around client confidentiality that overlay the Privacy Act requirements. Sending client financial records to an overseas AI platform without appropriate contractual protections is a risk in this context.

Legal professionals. Legal professional privilege and confidentiality obligations are separate from the Privacy Act but interact with it. Solicitors and barristers in particular need to consider whether inputting client matter information into a cloud-based AI tool is consistent with their professional obligations, regardless of what the Privacy Act requires.

Businesses contracting to government. If your business provides services to Commonwealth or state government agencies, you may be subject to additional privacy obligations under government procurement requirements, including requirements about where data is stored and processed.

What to Actually Do Next

The APPs are not a reason to stop using AI tools. They are a framework for using those tools responsibly, and most of what they require is practical and achievable for a business without a dedicated legal team.

A reasonable starting point for most SMBs is this: list every AI tool your business currently uses, identify which ones process personal information, and work through the six checklists above for each one. That audit will surface your actual gaps, as distinct from the theoretical ones, and give you a prioritised list of things to address.

Our free AI Compliance Checker tool walks through many of the same questions covered in the six checklists above and gives you a quick read on where your gaps are.

For a deeper read on the OAIC's own guidance on AI and privacy, including how the regulator expects businesses to interpret the APPs in the context of AI adoption, see our OAIC AI guidance breakdown. For the specific question of whether it is safe to enter customer data into ChatGPT, the customer data and ChatGPT article works through the details.

Related reading: our Claude AI review for Australian business.

Try our free AI Privacy Risk Scorer to score your current AI tool setup against Privacy Act requirements.

Try our free AI Compliance Checker to check whether your AI tools meet your compliance obligations.

Related reading: our AI governance by region.

Do the Australian Privacy Principles apply to my small business if I use AI tools?

The Privacy Act 1988 and the APPs apply to your business if your annual turnover exceeds $3 million, or if you operate in certain industries regardless of turnover (including health services, credit providers, and government contractors). If you meet those thresholds, the APPs apply to everything you do with personal information, including when an AI tool is doing the processing on your behalf. If you're below the threshold and not in a covered industry, the Act may not apply, but you may still face contractual or reputational obligations to handle customer data responsibly.

What counts as personal information under the Privacy Act?

Personal information is any information or opinion about an identified individual, or an individual who is reasonably identifiable. This includes names, email addresses, phone numbers, home addresses, health information, financial details, and in many cases IP addresses and device identifiers. It does not need to be sensitive or secret to count as personal information. If it can identify a specific person, it is personal information for the purposes of the Act.

Can I use overseas AI tools like ChatGPT without breaching APP 8?

Yes, but you need to address APP 8 rather than ignore it. The most common approaches are: signing a data processing agreement with the vendor that includes APP-equivalent protections, relying on the individual's informed consent to the overseas transfer, or verifying that the overseas country has comparable privacy protections. Simply using an overseas AI tool without taking any of these steps is the gap most SMBs currently have. The customer data and ChatGPT guide works through this specifically for OpenAI's platform.

What is the difference between APP 5 and APP 1, and do I need to address both?

APP 1 requires your business to have a Privacy Policy that accurately reflects how you manage personal information overall. APP 5 requires you to notify individuals at the point of collection about specific things: who you are, what you'll use their information for, and whether you'll disclose it overseas. Both apply when you use AI tools. APP 1 is the policy-level obligation; APP 5 is the point-of-contact notification. You need both: a current Privacy Policy that mentions AI tool use, and a collection notice or disclosure that appears when you actually collect personal information through an AI tool.

Does it matter if an AI tool uses my customer data to train its models?

Yes, it matters under APP 6. APP 6 limits how you can use or disclose personal information beyond the purpose for which it was collected. If an AI vendor uses data you provide to train their AI models, that constitutes a further use of the personal information. Whether this is permitted depends on the vendor's terms and whether your customers were notified and consented. Many major AI vendors now offer options to opt out of training data use, or provide enterprise tiers where training is excluded. Check your vendor's current terms rather than assuming a default position either way.

What happens if I don't comply with the Australian Privacy Principles?

The OAIC can investigate complaints and, for serious or repeated breaches, can take enforcement action including civil penalty proceedings. Fines for serious breaches can reach $50 million for corporations (or a formula based on turnover or benefit obtained if higher). Beyond regulatory penalties, a privacy breach can damage client relationships in industries where confidentiality is a core part of the service offering, such as accounting, legal, and health. Most SMBs face the compliance risk as a reputational one first and a regulatory one second, but both are real.