Does the Privacy Act cover our organisation?
The OAIC explains that a small business with annual turnover of $3 million or less is generally not covered by the Privacy Act on size alone, and lists kinds of small business that are covered regardless of turnover, including those below. If your organisation is covered, everything in this guide applies to staff use of AI tools. If it is not, the guidance still describes sensible practice, and you may have confidentiality duties under contracts or professional rules that apply anyway. Check your position against the OAIC's small business guidance or with a qualified adviser.
- Health service providers.
- Businesses that trade in personal information.
- Contracted service providers under a Commonwealth contract.
- Small businesses that have chosen to opt in.
What does the OAIC say about staff using AI tools?
The OAIC's guidance on privacy and the use of commercially available AI products, published in October 2024 and updated in January 2025, opens with five takeaways. Privacy obligations apply to personal information put into an AI system and to output that contains personal information. Organisations should do due diligence before adopting a product. APP 6 limits using information for purposes other than the one it was collected for. Privacy policies and notices should explain the organisation's use of AI. And as best practice, organisations should not enter personal information, especially sensitive information, into publicly available generative AI tools.
When is pasting information into an AI tool a secondary use?
Under APP 6, an organisation may use or disclose personal information only for the purpose it was collected for, unless the person consents, or would reasonably expect the secondary use and it is related to the primary purpose, or directly related for sensitive information. The OAIC notes a secondary use may fall within reasonable expectations if it was set out in a collection notice and the privacy policy. For staff, the practical test is whether the customer, patient or employee would expect their details to go into that tool for that task.
Hypothetical example: a staff member pastes a customer's complaint, with name, address and account history, into a free chatbot to draft a reply. Removing the identifying details first, or using the organisation's approved business tool under a rule that permits it, changes the privacy position. Which option is acceptable is for your organisation to decide with its privacy adviser.
Does it matter if the AI output about a person is wrong?
Yes. The OAIC states that inferred, incorrect or artificially generated information about an identified or reasonably identifiable person, including hallucinations, is personal information and must be handled under the APPs. Generating or inferring personal information with AI counts as a collection under APP 3. APP 10 requires reasonable steps to keep personal information accurate, and the OAIC suggests records should show where information came from an AI output, with a person responsible for checking it. For staff, an AI summary of a client's history is a draft until someone has checked it against the source.
What about information about employees?
A private sector employer's handling of employee records is exempt from the APPs when the act or practice is directly related to a current or former employment relationship. The OAIC explains the limits. The exemption does not cover future employment relationships, so it does not apply to job applicants who are not employed. It does not cover contractors handling another organisation's employee information, and it is unlikely to apply to organisations providing recruitment, HR, training or similar services under contract. HR teams using AI on applications should check their position before relying on the exemption.
What should staff training cover?
The OAIC says that controls or prohibitions on including personal information in AI inputs need to be accompanied by training and auditing measures, and that staff training should continue through the life of an AI product. A session should make the rule practical: which tools are approved, which kinds of information never go in, how to remove identifying details when a task allows it, what to check in an output about a person, and what to do after a mistake.
- Practise on fictional material and check it first: the OAIC describes a chatbot asked for fictional case studies that produced a real case with real names.
- Show the difference between a business account and a consumer account in your chosen tool.
- Name the person staff contact if they have entered something they should not have.