What are the Australian Privacy Principles?
The Australian Privacy Principles are the 13 principles that sit in Schedule 1 of the Privacy Act 1988. They are the operative privacy rules for entities the Act covers, and the OAIC publishes them in a quick reference that states each principle in a single line.
In the OAIC's own summaries, APP 1 ensures that APP entities manage personal information in an open and transparent way. APP 2 requires entities to give individuals the option of not identifying themselves, or of using a pseudonym. APP 3 outlines when an entity can collect personal information that is solicited, and applies higher standards to the collection of sensitive information. APP 4 outlines how entities must deal with unsolicited personal information. APP 5 outlines when and in what circumstances an entity that collects personal information must tell an individual about certain matters.
APP 6 outlines the circumstances in which an entity may use or disclose personal information that it holds. APP 7 permits use or disclosure for direct marketing only if certain conditions are met. APP 8 outlines the steps an entity must take to protect personal information before it is disclosed overseas. APP 9 outlines the limited circumstances when an organisation may adopt a government related identifier as its own.
APP 10 requires reasonable steps to ensure personal information collected is accurate, up to date and complete. APP 11 requires reasonable steps to protect personal information from misuse, interference and loss. APP 12 sets out obligations when an individual requests access to personal information held about them, and APP 13 sets out obligations to correct it.
Read in sequence, the principles trace a lifecycle: govern it, collect it lawfully, tell people, limit what you do with it, protect it, and let people see and fix it.
Who do the Australian Privacy Principles apply to?
The APPs bind APP entities. That means most Australian Government agencies and private sector organisations with an annual turnover of more than $3 million.
Turnover is not the whole test. The OAIC lists small business operators that are covered regardless of size, including private sector health service providers, businesses that sell or purchase personal information, credit reporting bodies, contracted service providers under an Australian Government contract, accredited participants in the Consumer Data Right, and businesses that have opted in to the Act. Conversely, state and territory agencies, public schools, individuals acting in a personal capacity, and registered political parties generally sit outside.
Where do the Australian Privacy Principles fit with AI?
The OAIC has published guidance for organisations using commercially available AI products, and its framing is that privacy obligations apply to any personal information input into an AI system as well as the output data generated by AI where that output contains personal information.
Four principles carry most of the weight. APP 3 is engaged when an AI system generates or infers personal information, because the OAIC treats that as a collection, and it states that inferred, incorrect or artificially generated information produced by AI models, including hallucinations and deepfakes, is personal information where it concerns an identified or reasonably identifiable individual. APP 6 restricts what an entity can do with personal information it puts into an AI system, limiting use to the primary purpose unless there is consent or the secondary use is both reasonably expected and related. APP 8 is triggered by hosted models running offshore. APP 11 requires reasonable security steps over both prompts and outputs.
APP 1 is the governance anchor. The OAIC says organisations should update privacy policies and notifications with clear and transparent information about their use of AI, including ensuring that public-facing AI tools such as chatbots are clearly identified as such. From 10 December 2026 the APP 1 policy obligation extends to describing automated decision making that could significantly affect an individual's rights or interests.
What should practitioners do under the Australian Privacy Principles?
Map each AI use case to the specific principles it engages rather than treating privacy as one undifferentiated obligation. A retrieval tool over internal documents is largely an APP 6 and APP 11 question. A hosted model is also an APP 8 question. A system that scores or ranks people is an APP 1 and APP 3 question as well.
The OAIC recommends a privacy by design approach including a privacy impact assessment, due diligence before adoption that considers whether the product has been tested for the intended use and how human oversight can be embedded, and regular review rather than a set and forget posture. It also recommends organisations do not enter personal information, particularly sensitive information, into publicly available generative AI tools, which is the cheapest control most organisations can implement immediately.
For HR and people teams the practical trap is assuming workforce data is out of scope. Confirm the position before running analytics, and read this alongside the Fair Work Act 2009. Related material sits under the privacy topic hub.
Bottom line
The Australian Privacy Principles are the operative privacy rules for the entities the Privacy Act covers, and read in sequence they trace a lifecycle: govern it, collect it lawfully, tell people, limit what you do with it, protect it, and let people see and fix it. They apply to AI without amendment, because the OAIC's position is that obligations attach to any personal information put into a system as well as to output the system generates where that output concerns an identifiable person. The practical task is therefore mapping each use case to the principles it actually engages, since a retrieval tool over internal documents raises different questions from a hosted model or a system that scores and ranks people. For Australian practitioners the cheapest control remains the plainest one, keeping personal information and particularly sensitive information out of publicly available generative AI tools, and the trap worth naming is assuming workforce data sits outside the Act.
TheAICommand. Intelligence, At Your Command.*
TheAICommand. Intelligence, At Your Command.