AI policy and privacy
Can you put customer information into AI tools?
Usually not, if the tool is a public one. The OAIC recommends that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. That is guidance, not a flat legal ban, but the Privacy Act still applies to how personal information is used and disclosed.
Does the Privacy Act apply to my workplace?
The Privacy Act 1988 covers Australian Government agencies, organisations with annual turnover over $3 million, and some other organisations. These are called APP entities. They follow the 13 Australian Privacy Principles, or APPs.
If your employer is an APP entity, the privacy rules reach the way your team handles customer information, including in AI tools. The OAIC's AI guidance is written for organisations that use AI, and it covers internal use as well as public-facing tools.
Not sure if your employer is covered? Ask your manager or privacy contact. Don't assume you are exempt.
What does the OAIC say about public AI tools?
In October 2024 the OAIC published its guidance on privacy and the use of commercially available AI products. It covers freely available tools such as public chatbots.
The key advice is a best-practice recommendation. Don't enter personal information, especially sensitive information, into publicly available generative AI tools. The OAIC gives the reason as significant and complex privacy risks.
A December 2025 OAIC blog repeats the advice. It adds that once personal information goes in, it is very difficult to track or control how it is used, and it may be impossible to remove.
The OAIC also says its guidance is not a full statement of every AI privacy issue.
Which privacy principles matter most?
The OAIC points to several APPs. In plain terms they say this.
- APP 6 (use and disclosure). Entering personal information into an AI tool can be a use or a disclosure. Where the tool's provider can access what you enter, the OAIC treats it as a disclosure to the provider. Using information for a purpose other than the one it was collected for generally needs the customer's consent, or it must be something they would reasonably expect.
- APP 8 (overseas disclosure). Cloud tools may keep data on servers outside Australia, which can be an overseas disclosure.
- APP 11 (security). Organisations must take reasonable steps to protect personal information they hold.
- APP 10 (accuracy). AI outputs can be wrong. The OAIC says not to record or treat them as facts, and a person should check the accuracy of any personal information they contain.
- APP 3 (collection). If AI generates or infers personal information about someone, that counts as collection.
- APP 1 (openness). Organisations must have practices that support compliance and a current privacy policy.
The OAIC also notes that made-up or incorrect content about a person who can be identified, such as a hallucination or a deepfake, can still be personal information.
What can go wrong when staff paste in customer details?
The OAIC's December 2025 blog uses a made-up example to show the risk. In it, an employee uploads a customer's financial hardship application to a public AI chatbot to get a summary. The application includes health and family details. The business's own policy bans this.
In the example, the summary leaves out key details and the application is refused. The customer suffers significant financial and emotional distress.
The OAIC's point is that controls alone are not enough. Staff need to know what can and cannot be entered into an AI platform.
Is Australian privacy law changing?
One change is confirmed and has a date. The Privacy and Other Legislation Amendment Act 2024 added a transparency duty to APP 1. From 10 December 2026, APP entities must include certain information in their privacy policy about some automated decisions.
It applies where an entity has arranged for a computer program to make, or do something substantially and directly related to making, a decision. The decision must reasonably be expected to significantly affect a person's rights or interests, and the program must use that person's personal information. The OAIC published fact sheets on this in September 2026.
This mostly affects organisations, not individual staff. But it is a good reason to know where your team uses AI.
What should I do and not do?
Do
- Check your organisation's AI policy before you use any AI tool with customer details.
- Use tools your organisation has approved. The OAIC says to check vendor terms on whether inputs are used for training.
- Leave out names, contact details, account numbers, and health or financial information. Describe the situation in general terms instead.
- Check every AI output before it goes to a customer or into a record.
- Ask your privacy contact when you are unsure.
Don't
- Paste customer files, emails or applications into a public chatbot.
- Treat AI output about a person as fact.
- Assume a tool is private because it feels like a one-to-one chat.
What can I do this week?
- Find your organisation's AI policy and read the part on personal information.
- Ask who your privacy contact is.
- List the AI tools you use at work and note which are approved.
- Take one task where you paste in customer details. Redo it with the details removed.
- Check whether your team knows what it can and cannot enter.
Want a clear, practical start? AI Work Ready's free course, Responsible AI for Australian Workplaces, is at aiworkready.com.au. Teams can see more on the organisations page.
This is general information, not legal advice. Check your organisation's AI policy.
Sources
- Guidance on privacy and the use of commercially available AI products, OAIC, 21 October 2024
- GenAI tools in the workplace: balancing protection of personal information and business efficiency, OAIC, 4 December 2025
- Top five takeaways: Privacy considerations when using commercially available AI products, OAIC
- New resources on transparency for use of AI and automated decision-making, OAIC, 30 September 2026
- The Privacy Act, OAIC
Learn the habit, free
Responsible AI for Australian Workplaces teaches what never goes into an AI tool, how to ask well and how to check the answer. Short online lessons, free for individuals.