Australia's New AI Transparency Rules Start 10 December: What Businesses Using AI Must Do

Short answer: From 10 December 2026, organisations covered by the Privacy Act must say in their privacy policy when a computer program uses personal information to make, or substantially help make, decisions that could significantly affect people. That includes AI agents, scoring tools and even spreadsheet formulas. The policy must name the kinds of personal information used and the kinds of decisions involved, in plain language.

If your business uses AI or automation to sort leads, approve applications, set prices, triage enquiries or screen job candidates, this change applies to the way you describe that work. It isn’t a ban on automation. It’s a transparency rule, and the deadline is close.

What is changing on 10 December 2026?

Direct answer: new Australian Privacy Principles (APP 1.7 to 1.9) require your privacy policy to explain automated decision-making. They were introduced by the Privacy and Other Legislation Amendment Act 2024 (Cth) and take effect on 10 December 2026.

The obligation sits with the business that arranged for the program to be used, not with the software vendor. If you use a third-party AI tool to make or guide decisions about customers, the disclosure is yours to make.

Does it apply to my business?

Direct answer: it applies to organisations covered by the Privacy Act. Most businesses with an annual turnover of $3 million or less are exempt from the Act, but some are covered regardless of size, including health service providers and businesses that trade in personal information. If you already have to follow the Australian Privacy Principles, assume this applies to you.

What counts as an automated decision?

The rule covers decisions that could reasonably be expected to significantly affect a person’s rights or interests. The effect can be good or bad. Examples given in commentary on the law include:

  • refusing or approving access to significant services such as credit, banking, healthcare or utilities
  • decisions about contract rights, such as insurance
  • differential pricing for different customers
  • decisions that limit employment opportunities, such as automated candidate screening

A human in the loop doesn’t take you outside the rule. If the program recommends or guides the person who makes the final call, and its work is substantially and directly related to the decision, it counts. The guidance even uses a spreadsheet formula that triages hotline calls as an example of something that would be covered.

What must the privacy policy say?

In plain language, your privacy policy needs to set out:

  1. the kinds of personal information used in the program
  2. the kinds of decisions made solely by the program
  3. the kinds of decisions where the program does something substantially and directly related to the decision

Generic wording such as “we use technology to improve our services” won’t meet the standard. You don’t have to reveal commercially sensitive details of how your systems work.

A five-step plan before 10 December

  1. List every tool that touches a decision. Include AI agents, CRM lead scoring, chatbots, pricing rules, form-based approvals and spreadsheets, plus any third-party software doing this for you.
  2. Apply the “significantly affect” test. Which of those decisions could materially change what a customer, applicant or employee gets?
  3. Map the personal information. For each covered decision, note what information goes in and what the output is used for.
  4. Update the privacy policy. Add a plain-English section covering the three points above, and have it checked by your lawyer.
  5. Keep it current. Add a review step whenever you introduce or change an automation. A set-and-forget approach won’t hold up.

What this means if you’re building AI agents

Well-built automation already documents what each step does and which data it uses, so meeting this rule is mostly a matter of writing it down clearly. When we design process automation or an AI receptionist for a client, we map the data flow and decision points as part of the build, which gives you what you need for the privacy policy update. For the bigger picture of where agents fit, see AI agents vs marketing automation.

This article is general information, not legal advice. Speak to your lawyer about your own privacy policy.

Frequently asked questions

When do the automated decision-making rules start?

On 10 December 2026, under changes made by the Privacy and Other Legislation Amendment Act 2024.

Does using ChatGPT or an AI chatbot count?

It depends on what the tool decides. If it uses personal information to make or substantially guide a decision that could significantly affect someone, such as approving, pricing or screening, it is likely covered. Drafting a marketing email generally isn’t a decision about a person.

If a staff member makes the final decision, am I covered?

Possibly. A human in the loop doesn’t exempt you if the program’s recommendation is substantially and directly related to the decision.

Who is responsible when the AI tool comes from a vendor?

The business that arranged to use the program in its decision-making, not the software vendor.

Do I have to explain how my algorithm works?

No. You must describe the kinds of personal information and the kinds of decisions in plain language. Commercially confidential details of the system don’t have to be disclosed.

Sources: Corrs Chambers Westgarth; Hamilton Locke.

Write A Comment

Your email address will not be published. Required fields are marked *

    Are you looking for a Marketing Partner?

    Get in touch today and start achieving your business objectives.

    Prefer to talk? Book a time with Richard or call 1800 512 124.