What Australian businesses need to understand about AI tools, data security, and governance — and how to build a cyber security posture that keeps pace with how your team is already working
AI tools are already part of how Australian businesses operate. Staff are using ChatGPT, Claude, Microsoft Copilot, and Gemini to draft documents, summarise meetings, analyse data, and handle tasks that would otherwise take hours.
Most of this is happening without a formal policy, without documented controls, and without a clear picture of where business data goes once it leaves the screen.
That's not a reason to stop using AI tools. It's a reason to understand them properly — and to put the governance in place that protects your business, your clients, and your compliance obligations while your team uses them.
This guide covers what Australian businesses need to know about AI and cyber security in 2026 and beyond.
This is the question most businesses haven't asked — and it's the most important one.
When a staff member enters information into a public AI tool, the data doesn't stay on their device. It follows a path: from your device, to the AI tool's servers, and often to external infrastructure — frequently hosted outside Australia — where it may be retained for a period defined by the provider's terms of service, not by your organisation's data handling policies.
The Office of the Australian Information Commissioner (OAIC) published guidance in October 2024 making clear that the Privacy Act 1988 and the Australian Privacy Principles apply to any use of AI involving personal information — including internal business use of commercially available AI tools like chatbots, content-generation platforms, and productivity assistants.
The practical implication: when a staff member pastes client information, financial records, or business correspondence into a public AI tool, your organisation may be responsible for what happens to that data — regardless of whether you authorised the action or not.
According to the University of Melbourne and KPMG's Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025, 70% of Australian employees are using free, public AI tools rather than employer-provided, governed versions. That figure represents a significant and largely unmanaged data exposure for most Australian businesses.
Most Australian businesses don't have a formal AI usage policy. The same University of Melbourne and KPMG study found that 68% of Australian employees report their organisation has no policy on generative AI use.
Without a policy, what data gets shared with AI tools is left entirely to the individual behind the keyboard. Different staff make different decisions. Different AI tools operate under different terms. And there is no documentation to show an insurer, an auditor, or a regulator if something goes wrong.
This matters for several reasons that are now formally recognised in Australian law and regulation:
The Cyber Security Act 2024 formalises the expectation that cyber risk — including the risks introduced by AI tool adoption — is managed and governed at a leadership level, with documentation to support it. Businesses that cannot demonstrate they have considered and managed AI-related data risk are increasingly exposed under this framework. For a full breakdown of what the Act requires, see our post on what business leaders are now expected to demonstrate under the Cyber Security Act 2024.
The OAIC's AI guidance makes clear that organisations must conduct due diligence before adopting AI tools, document how personal information flows through those tools, and ensure their privacy policies accurately reflect AI usage. The OAIC recommends organisations do not enter personal information into publicly available AI platforms due to the significant privacy risks involved.
The Privacy Act 1988 applies to AI-related data handling regardless of whether a business intended to use AI for that purpose. If personal information enters an AI system, the Australian Privacy Principles apply to how it is collected, used, stored, and disclosed.
The absence of an AI policy is not a neutral position. It is a governance gap with growing legal and commercial consequences.
Consider two Australian businesses at the same point in 2024. Both have staff using AI tools daily. Both are at the same size and in the same industry. The difference is how they responded to that reality.
Business A — Reactive
Business B — Proactive
The starting point was identical. The outcome was not. The difference was governance — not technology.
Not all AI tools handle data the same way. The distinction that matters most for Australian businesses is between public AI tools and enterprise or governed versions of the same tools.
Public AI tools (such as the free tiers of ChatGPT, Claude, and Gemini) process data on external servers, may retain it for model training depending on the provider's current terms, and do not operate within your organisation's security or compliance boundary.
Enterprise and governed versions (such as Microsoft 365 Copilot, Google Workspace with Gemini, or Claude for Enterprise with data processing agreements) operate under contractual data handling terms that address retention, processing location, and training use. They do not automatically mean data is safe — but they provide a framework of accountability that public tools do not.
Understanding which tools your staff are using — and which category they fall into — is the starting point for any meaningful AI governance program.
The second shift is moving from a binary "AI is allowed / AI is not allowed" position to a documented, specific policy that covers:
This doesn't need to be a lengthy document. It needs to be clear, communicated, and documented. The OAIC's guidance on privacy and the use of commercially available AI products provides a useful framework for thinking through these questions.
Tools like Microsoft 365 Copilot and Google Workspace Gemini are integrated into your business environment in a way that public AI tools are not. They can read across Outlook, SharePoint, and Teams (in the case of Copilot), or Gmail, Drive, and Docs (in the case of Gemini).
This is both the value proposition and the governance consideration. These tools can produce outputs informed by your actual business data — which makes them significantly more useful than a public AI tool working only from what you type into a prompt. But it also means the access controls, permission structures, and data governance in your Microsoft 365 or Google Workspace environment directly determine what these tools can see and use.
The ASD's Essential Eight — particularly the controls around restricting administrative privileges, application control, and Microsoft Office macro settings — provide the technical foundation for managing this access appropriately. An organisation with well-implemented Essential Eight controls has a meaningful governance layer over what AI tools can access within their environment.
A functional AI cyber security policy for an Australian business doesn't need to be complex. It needs to cover the following minimum elements:
Approved tools list: A documented list of AI tools approved for business use, differentiated by use case where appropriate. Updated as new tools emerge or existing tools change their terms.
Data classification guidance: Clear guidance on what categories of data are appropriate for AI tools, differentiated by tool type. At minimum: what is never appropriate (client personal information, sensitive financial data, confidential business records) regardless of tool, and what is appropriate only in governed, enterprise tools.
Staff awareness: Staff need to understand the policy, why it exists, and what to do if they are unsure. A policy that exists only as a document is not a governance control.
Incident response: A defined process for what happens if data is inadvertently shared with an AI tool inappropriately — who is notified, what steps are taken, and whether the incident triggers notification obligations under the Notifiable Data Breaches scheme.
Review cadence: AI tools, their terms of service, and the regulatory environment around them are changing rapidly. The policy needs a defined review cycle — at minimum annually, and whenever a significant new tool is adopted or existing terms change materially.
The Essential Eight maturity model is not an AI-specific framework. It was designed to protect Australian organisations against the most common cyber threats. But several of its controls are directly relevant to managing the risks introduced by AI tool adoption.
Restricting administrative privileges: limits the data that any AI tool integrated into your environment can access — including governed tools like Copilot and Gemini — by ensuring access controls are applied correctly.
Application control: gives organisations visibility and control over which applications — including AI tools — can run in the business environment, reducing the risk of unsanctioned tools being used with business data.
Patching and configuration management: ensures that the enterprise AI tools your business does use are kept current and correctly configured, reducing the risk of vulnerabilities in those tools being exploited.
Multi-factor authentication: protects the accounts that AI tools operate under — if an account used to access Copilot or Gemini is compromised, the AI tool's access to business data is compromised with it.
For businesses that have not yet assessed their Essential Eight maturity, understanding where controls sit — and where gaps exist — is the foundation of a credible AI governance posture. A business that cannot demonstrate its Essential Eight maturity cannot demonstrate that it has managed the cyber security risks associated with AI tool use.
An Essential Eight assessment is the starting point for that picture.
The governance requirements around AI in the workplace are not separate from the cyber security and compliance program DefenderSuite is built around. They are an extension of it.
DefenderSuite supports AI governance through:
Essential Eight alignment and assessment: Mapping controls to the ASD maturity model — including the controls most directly relevant to AI tool access and data governance — and producing the documented maturity score that underpins any defensible governance position. Book a free Essential Eight assessment to understand your current position.
Microsoft 365 security governance: Configuring and managing the security controls in Microsoft 365 — including access management, sensitivity labelling, and compliance policies — that determine what Copilot and other integrated AI tools can access within your environment.
Compliance documentation: Producing the ongoing evidence — monthly security reports, access control logs, patching records — that demonstrates to insurers, auditors, and clients that AI-related risks are being actively managed, not just acknowledged.
Incident response readiness: Ensuring businesses have a defined, documented response process for data incidents — including those that involve AI tools — consistent with the notification obligations under the Notifiable Data Breaches scheme and the Cyber Security Act 2024.
AI governance is not a separate program from cyber security governance. It is part of the same question: does your business have the controls, the documentation, and the oversight structure to demonstrate that it is managing its data responsibly?
For most Australian businesses, the honest answer to that question requires a starting point — a precise, evidence-based picture of where controls currently sit, where the gaps are, and what it would take to reach a position that holds up under scrutiny.
Book your free Essential Eight assessment with DefenderSuite
Assessment completed within 5 business days. Included with a DefenderSuite 12-month plan. Minimum 10 users.
Related reading:
Sources:
University of Melbourne and KPMG – Trust, Attitudes and Use of Artificial Intelligence: A Global Study 2025: https://kpmg.com/au/en/insights/artificial-intelligence-ai/trust-in-ai-global-insights-2025.html
University of Melbourne – Full research report (Figshare): https://figshare.unimelb.edu.au/articles/report/Trust_attitudes_and_use_of_artificial_intelligence_A_global_study_2025/28822919
OAIC – Guidance on privacy and the use of commercially available AI products: https://www.oaic.gov.au/privacy/privacy-guidance-for-organisations-and-government-agencies/guidance-on-privacy-and-the-use-of-commercially-available-ai-products
OAIC – Notifiable Data Breaches scheme: https://www.oaic.gov.au/privacy/notifiable-data-breaches
ASD – Essential Eight Explained: https://www.cyber.gov.au/business-government/asds-cyber-security-frameworks/essential-eight/essential-eight-explained
ASD – Essential Eight Maturity Model: https://www.cyber.gov.au/business-government/asds-cyber-security-frameworks/essential-eight/essential-eight-maturity-model
Privacy Act 1988 (Cth): https://www.legislation.gov.au/Details/C2022C00361