At the same time, the Australian Government has raised its own standard. Version 2.0 of its Policy for the responsible use of AI in government took effect on 15 December 2025. It sets clearer expectations for governance, accountability and transparency when agencies use AI, including in recruitment. Agencies must assess their in-scope AI uses, and review existing ones by 30 April 2027. 

If you hire for the public sector, this applies to you now. If you supply to government, or want to, it is the benchmark buyers will hold you to. 

This is the point most teams miss. People treat “AI” and “automated decision making” as the same idea. They are not, and the difference decides which rules apply. 

Automated decision making, or ADM, is any use of a computer program to automate part of a decision. A cut-off score that progresses candidates is ADM. 

AI is one method that can be used inside ADM. It is not the only one. Most automated scoring in assessment runs on fixed rules, not AI. 

The Privacy Act’s transparency rules apply to ADM. The Government’s policy applies to AI. To meet both, you have to know which is which in your own process. You cannot disclose or explain a step you have never mapped. 

What a defensible process looks like 

Being defensible does not mean removing automation from hiring. It means being able to stand behind your decisions when someone questions them. Five things make the difference. 

1. You know where AI and ADM sit. You have mapped your recruitment and assessment tools, and you can say which steps are automated and which are not. 

1. A person owns each decision. Automated results inform the decision. They do not make it. A named person is accountable. 

1. The logic is clear and stable. You can explain how a score or recommendation is produced, and that logic does not change over time without anyone noticing. 

1. You disclose it. Your privacy policy matches what you actually do, and it is updated before December 2026. 

1. You govern it. Scoring models and reference data are checked, results are monitored for fairness and consistency, and every automated feature has a policy and a review behind it. 

You have time to do this properly. A practical starting point: 

• Audit your hiring technology. Find every step that uses personal information to make or support a decision about a candidate. 

• Question your vendors directly. Does the tool use AI to score, rank, shortlist or reject candidates? Does the scoring change over time? Where is candidate data stored? What security certifications does the vendor hold? If a vendor cannot answer, that risk becomes yours. 

• Review your privacy policy against what you actually do, and plan the update now rather than close to the deadline. 

• Record your human oversight. Write down who reviews what, and where a person signs off. 

The opportunity 

Good technology still belongs in hiring. Assessment science, automation and data, the foundations of Hiring Intelligence, help you hire faster and more consistently than judgement alone, and they reduce bias when they are built and used well. The new rules do not ask you to slow down. They ask you to show your work. 

For employers who chose their tools with care, that is straightforward. For those who did not, it is a real risk. It comes down to one moment. When a hiring decision is challenged, can you give a clear, documented answer that a person stands behind? 

Testgrid was built around that question. Assessments are scored using established psychometric methods, not AI. Automated steps stay transparent and under the customer’s control. Candidate data is held in Australia. Disclosure is built in. If you are working through what these reforms mean for your hiring, we are happy to show you how we have approached it.