The OAIC’s AI guidance: what it asks of a business that uses AI tools.
In short: on 21 October 2024 the Office of the Australian Information Commissioner (OAIC) published two guides on AI and privacy. The one most businesses need is its Guidance on privacy and the use of commercially available AI products, which explains how the Privacy Act 1988 and the Australian Privacy Principles (APPs) apply when an organisation selects and uses an AI product. Its best-known line recommends against entering personal information into publicly available generative AI tools.
This page sets out what the guidance is, who it is for, what it says principle by principle, and where the choice of deployment changes the analysis.
General information only. This guide is not legal advice and is not a substitute for advice about your specific obligations. Verify positions against the primary sources linked throughout, and take professional advice where it matters.
What the guidance is, and who it is for
The OAIC says the guidance is “intended to assist organisations to comply with their privacy obligations when using commercially available AI products, and to assist with selection of an appropriate product”. It is aimed at any organisation using AI to provide a product or service, including internally, and it also covers freely available products “such as publicly accessible AI chatbots”.
The obligations themselves come from the Privacy Act and the APPs, which the OAIC says “apply to all uses of AI involving personal information”. It is not a comprehensive overview, and is to be read with the Act and the APP guidelines.
Current version, and where to read it
The guidance was published on 21 October 2024, and the OAIC page shows it as updated on 17 January 2025. It is a single web page with five takeaways, a quick reference guide, worked examples and two checklists, which also cover internal AI policies, monitoring and staff training.
Its companion, the Guidance on privacy and developing and training generative AI models, was published the same day and is shown as updated on 23 October 2024. It is written for developers, and its headline point is that data being publicly available “does not mean it can legally be used to train or fine-tune generative AI models or systems”.
What it says, principle by principle
Organisations “should conduct due diligence to ensure the product is suitable to its intended uses”, including whether it has been tested for those uses and who will have access to personal information input or generated.
Businesses should update their privacy policies and notifications “with clear and transparent information about their use of AI”, and identify public-facing AI tools such as chatbots as AI.
“If AI systems are used to generate or infer personal information, including images, this is a collection of personal information and must comply with APP 3.” That includes incorrect output such as hallucinations, where it is about an identified or reasonably identifiable individual.
Personal information put into an AI system may only be used or disclosed for the primary purpose it was collected for, unless the individual has consented or would reasonably expect the secondary use and it is related to the primary purpose.
AI systems “are known to produce inaccurate or false results”. Organisations must take reasonable steps to ensure accuracy, with a human user responsible for verifying any personal information obtained through AI.
Consider where servers are located and whether personal information could be disclosed outside Australia. Our APP 8 guide covers the cross-border rules.
The recommendation on publicly available AI tools
The fifth takeaway reads: “As a matter of best practice, the OAIC recommends that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools, due to the significant and complex privacy risks involved.” The guidance explains why: once personal information has been input into an AI system it will be very difficult to track or control how it is used, “and potentially impossible to remove the information from the system”.
How it applies when a business uses AI tools
Entering personal information into an AI system may be a use “if the data stays within your organisation’s control”, or a disclosure if it is made accessible to others outside the organisation and released from its effective control.
The OAIC gives two examples. In the first, employees at an insurance company enter a customer’s claim details, including health information, into a publicly available chatbot. The OAIC says the company “is disclosing the information to the owners of the chatbot”, and must show the disclosure fits the purpose of collection or an APP 6 exception. In the second, an organisation uses a proprietary AI system with protections so that information is not disclosed outside the organisation, which the OAIC says “will constitute a use rather than a disclosure”.
Where the developer of an AI system has access to personal information processed through it, the OAIC says that is a disclosure that should be included in an APP 5 notice. Our Privacy Act guide covers the principles in more detail, and the use cases series shows what an ordinary document carries.
Where deployment posture fits
Most of the guidance is about conduct, and no product does that work for an organisation: the due diligence, the notices, the accuracy checks and the training remain its own. The use and disclosure question depends on where information goes, and here the guidance is direct: “Deploying a system locally or on premises is likely to be more privacy-preserving as it limits the risks of third party access to the data.”
An AI unit inside the business processes documents on the organisation’s own hardware, so the information stays within its control and no developer receives it, which is closer to the situation the OAIC describes as a use. The OAIC does not assess particular products, and that judgement is the organisation’s to make. APP 6, APP 3 and APP 10 still apply. On Ermos Edge, documents stay on the organisation’s own instance in Australia and only permitted excerpts go, per question, to a named Australian provider under contract. The organisation still has to assess that data flow under APP 6 and describe it in its APP 5 notice and privacy policy.
Common questions
When was the OAIC AI guidance published, and has it been updated?
Can a business enter personal information into a public AI chatbot under the Privacy Act?
Is information generated by AI personal information?
Does a privacy policy need to mention the use of AI?
See Dominion answering from your own documents.
A demonstration takes 30 minutes, uses no client data, and comes with no obligation.
Book a demo