Monday–Friday, 9:00 AM–5:00 PM AEST · After-hours support available on request 1300 584 007
New AI transparency rules, deadline 10 December 2026 graphic

If your business uses any kind of software or AI tool to help make decisions about people — approving applications, screening candidates, scoring risk, prioritising customers — the Office of the Australian Information Commissioner wants your privacy policy to say so clearly, and it wants that done before 10 December 2026.

What the OAIC actually released

The OAIC has published new transparency guidance, including fact sheets and a decision-making flowchart, aimed at businesses that use AI or other automated systems to make, or substantially help make, decisions that affect a person’s rights or interests. The core requirement isn’t new in spirit — Australia’s Privacy Act has long required businesses to be clear about how personal information is used — but this guidance sharpens exactly what “clear” needs to look like when an algorithm, rather than a person, is doing some or all of the deciding. If that description matches any tool your business uses, your privacy policy needs to reflect it by the deadline.

Why this applies to more businesses than it sounds like

“Automated decision-making” sounds like it belongs to large tech platforms, but in practice it covers a lot of ordinary small and medium business software: a recruitment platform that auto-ranks applicants, a finance tool that auto-approves or flags transactions, a CRM that scores leads, or a tenant-screening service used by a real estate agency. None of these need to be exotic AI to count — if the system is doing meaningful work in a decision that affects someone, the transparency expectation applies. Working out which of your tools fall under this is a good job for an IT consulting review.

The practical risk of ignoring this isn’t abstract either. Privacy complaints and OAIC inquiries tend to start exactly where a business’s actual practice doesn’t match what its privacy policy claims — and “we didn’t realise our software counted as automated decision-making” is a weak position to explain after the fact, particularly once guidance specifically addressing it has been public for months.

What to do before December

1. List the software and AI tools your business actually uses that touch decisions about customers, applicants, or clients — not just the obviously “AI-branded” ones, but anything that scores, ranks, screens, or auto-approves. 2. For each one, check whether your current privacy policy actually mentions it, or whether it’s silent on automated processing entirely. 3. Update the policy wording to match the OAIC’s guidance before 10 December 2026 — this is specifically a wording and disclosure exercise, not a requirement to stop using the tools themselves.

This is squarely the kind of compliance detail that’s easy to get right with two months’ notice and genuinely awkward to explain after a complaint lands. If you’d like a second set of eyes on whether your current tools and policy actually line up, our cybersecurity and compliance team can review it with you well ahead of the deadline.

The OAIC’s full guidance, including the fact sheets and flowchart referenced above, is published directly on its transparency resources page.

Two months is enough time to do this properly and calmly. It’s not enough time to do it the week after someone asks why your policy doesn’t mention the tool that screened their application.

For a real example of AI access going wrong, see OpenAI’s unauthorised access to Australian government systems.

Not sure where your business stands? Book a free IT consultation.

Need a hand with this?

Get in touch and we'll walk you through it.