From 10 December 2026 You Must Disclose AI and Automation in Hiring

Dilara AlmeidaDilara Almeida18 November 20268 min read
Two HR colleagues talking across an empty meeting table in a bright Australian office

From 10 December 2026, automated decision-making hiring privacy rules in Australian Privacy Principles 1.7 to 1.9 require APP entities to say, in the privacy policy, when a computer program uses personal information to make a decision, or to do something substantially and directly related to making one, that could significantly affect a person. The OAIC published final automated decision guidance on 30 September 2026. Ranking software, scored assessments and AI video-interview outputs sit in the OAIC examples. A recruiter clicking approve at the end is not an automatic way out.

This page is the 10 December disclosure deadline, not a recap of the wider 2026 Privacy Act hiring changes. Use it to interrogate reference checks and assessments before you rewrite the policy.

Key Takeaways

  • APP 1.7 to 1.9 commence on 10 December 2026 and apply to decisions made from that date, including programs you arranged before then.
  • The OAIC treats recruitment ranking and AI video-interview scoring as in-scope examples where they use personal information and can block a job.
  • Human review does not automatically take a hiring tool out of APP 1.7 if the program’s output is a key factor with a direct connection to the decision.
  • Configuring a vendor product still counts as you arranging a computer program; the disclosure duty sits with the APP entity using the personal information.
  • Unsuccessful applicants are not covered by the employee records exemption, so candidate screening files stay inside the Australian Privacy Principles.

What automated decision-making hiring privacy disclosure requires

The Privacy and Other Legislation Amendment Act 2024 inserted APP 1.7, 1.8 and 1.9 into Australian Privacy Principle 1. They commence on 10 December 2026. From that date an APP entity’s privacy policy must carry extra information where all three APP 1.7 tests are met. The OAIC’s Chapter 1 guidelines (version 2.0), fact sheet and flowchart are the reading pack. This is a transparency duty. It does not, by itself, create a GDPR-style right to contest the hire.

If you already keep a hiring privacy page for AI recruitment disclosure in Australia, treat 10 December as a policy-content deadline. Point readers to the existing Privacy Act 2026 hiring changes note and keep this page on APP 1.7 to 1.9.

The three APP 1.7 tests in a hiring stack

APP 1.7 applies when all three limbs are true.

You arranged for a computer program

“Computer program” is wide. The OAIC fact sheet includes rule-based processes, AI and machine learning, ordinary software and apps, and generative AI including chatbots. A scoring formula can count. Procurement, configuration and customisation of off-the-shelf software are ways of arranging a program. Switching on ranking in an ATS, setting a cut-score on an assessment, or authorising staff to paste resumes into a chatbot that drafts a shortlist is arranging.

The same point applies to a vendor you pay. Chapter 1 examples include procuring another entity’s AI system to screen and rank job applications. The APP entity that uses the personal information to hire, or not hire, must speak in its own privacy policy. Vendor copy that says “we are the processor” does not move APP 1.7 off your letterhead.

The decision could significantly affect rights or interests

Hiring, promotion, pay and performance sit in the OAIC’s ordinary examples of decisions that can significantly affect a person. APP 1.9 adds that a decision includes refusing or failing to decide, and that effects count whether they help or harm the individual. Knocking someone out of a pipeline, never sending their file to a hiring manager, or leaving an application unanswered after a score can all be decisions. The fact sheet treats software that sorts candidate profiles, and a companion tool that scores a pre-recorded video interview (including voice, speech pattern and facial inputs), as capable of limiting access to employment.

Personal information runs the program

The third limb is that personal information about the individual is used in the operation of the program. Resumes, application answers, referee comments, assessment responses, video files and derived scores are personal information. Most candidate screening products are in that category.

Human-in-the-loop is not an automatic exemption

The usual objection is that a person still makes the offer. The OAIC fact sheet states that a decision can be in scope even where a computer program output does not replace the entire process or is subject to human review. “Substantially” means a key factor in the human’s decision. “Directly” means a direct connection to it. A recruiter who rubber-stamps a ranked list, or who only interviews people already scored above a threshold, is still using a program that is substantially and directly related to the decision.

Extensive human oversight and control can take some outputs out of scope. Rebuilding the shortlist from source files and recording why is a different fact pattern from a “human in the loop” checkbox. Record how often staff override the tool and whether they can proceed without the score.

Unsuccessful applicants are still in the APPs

The employee records exemption applies to acts directly related to a current or former employment relationship and an employee record. The OAIC is explicit: it does not cover future employment relationships. Unsuccessful applicants stay inside the APPs, including when a vendor holds their data for you.

What APP 1.8 requires you to write

If APP 1.7 is met, APP 1.8 says the privacy policy must describe the kinds of personal information used in those computer programs, the kinds of decisions made solely by those programs, and the kinds of decisions where a program does something substantially and directly related to making the decision.

Group similar processes, but stay specific enough that a reasonable applicant can tell what happens to them. “We may use AI” is not a kinds-of-decisions statement. Call out sensitive inputs such as biometric video analysis if you use them.

Keep ethics and model-governance questions on the pages already written for them: the AI hiring compliance framework and the ethical AI hiring guide. This deadline is the policy wording.

What to ask reference-check and assessment vendors before 10 December

The work is an inventory. For every reference-check and assessment product you configure, get written answers you can map to APP 1.8. The automated reference check compliance notes and the guide to AI recruitment privacy in Australia stay useful for vendor due diligence. Use them to extract the facts the policy now has to name, not as copy for the policy itself.

  • Outputs. Score, rank, pass/fail gate, recommended next step, or a narrative summary staff treat as a key factor?
  • Personal information. Identity, work history, referee answers, assessment items, video, audio, derived traits?
  • Who arranged the program. Which thresholds, knock-out rules and models did your team switch on?
  • Human role. Can a recruiter complete a hire without the output? How often is it overridden, and is that logged?
  • Sole versus related. Does any step run with no human in the path, for example an auto-reject below a cut-score?
  • Unsuccessful applicants. Where is rejected-candidate data held, for how long, and can you describe it for APP 1.8?
  • Policy wording. Can the vendor supply a plain-English description of kinds of information and kinds of decisions?

If a vendor cannot answer those questions, you cannot honestly complete APP 1.8. Existing tools still need a line in the policy on 10 December if they meet the three tests. Treat reference-check and assessment workflows as in-scope whenever personal information feeds a score, structured report or recommended next step staff rely on.

Related Resources

Common Questions About Automated Decision-Making in Hiring

1. When do APP 1.7 to 1.9 start?

APP 1.7 to 1.9 commence on 10 December 2026 and apply to decisions made from that date, even if you arranged the computer program or collected the personal information earlier.

2. Does a human signing off the hire take the tool out of APP 1.7?

No. Human review does not automatically take a process out of scope. If the program’s output is a key factor with a direct connection to the decision, APP 1.7 can still apply.

3. Do vendor ranking and assessment tools count as programs we arranged?

Yes, where you procure, configure or rely on the product in the hiring path. The APP entity using the personal information to decide who progresses is responsible for the privacy policy wording.

4. Are unsuccessful applicants covered by the employee records exemption?

No. The employee records exemption does not cover future employment relationships. Personal information about unsuccessful job applicants remains subject to the Australian Privacy Principles.

5. What must the privacy policy actually say?

APP 1.8 requires the kinds of personal information used in the programs, the kinds of decisions made solely by those programs, and the kinds of decisions the programs substantially and directly inform.

6. Is this the same as a right to contest an automated hire?

No. APP 1.7 to 1.9 are a privacy-policy transparency duty. They do not, by themselves, give a candidate a GDPR-style right to contest the decision.

Ask Vendors Before You Rewrite the Policy

You cannot describe kinds of information and kinds of decisions you have not mapped. Start with ranking, scored assessments, automated reference-check routing and any video-interview scoring you actually use. Get vendor answers in writing, then put APP 1.8 wording on the privacy policy by 10 December 2026.

AI Powered

Stop hiring by intuition.

Automate reference checks and skills assessments with Righteo. Get honest, structured insights on every candidate — faster and fairer. Trusted by 1,200+ Australian businesses.