With the automated decision-making (ADM) transparency obligation commencing on 10 December 2026, organisations are now in the final weeks of readiness. The timing is pointed. On 26 October 2026, the Office of the Australian Information Commissioner (OAIC) released its Notifiable Data Breaches (NDB) statistics for January to June 2026: notifications increased to 667 in this period, with malicious or criminal attacks remaining the leading cause. Large-scale breaches rose sharply. Breaches affecting 100,000 to 250,000 individuals doubled, and those affecting up to 1,000,000 individuals rose from zero to three in this period alone.

At the same time, the Australian Signals Directorate (ASD) continues to warn that frontier AI models are accelerating vulnerability discovery and exploitation, and that organisations that do not improve their defences will be particularly vulnerable to AI-enabled cyber threats. Clearly, the more personal information an organisation’s AI and decision-making systems draw on, the more there is to lose. ASD’s Annual Cyber Threat Report 2025-26, released on 27 October 2026, confirmed that the Australian Cyber Security Centre (ACSC) responded to more than 1,500 cyber security incidents and issued over 4,400 notifications of potentially malicious activity (almost an 80% increase) this year alone. It also found that AI is enabling malicious actors to operate with greater scale, speed and sophistication. Knowing, and being able to explain, where AI is used in your organisation is now both a compliance requirement and a clear security imperative.

On 30 September 2026, the OAIC released guidance on the new obligation, introduced by the Privacy and Other Legislation Amendment Act 2024 (Cth) (POLA Act) and contained in Australian Privacy Principle (APP) 1, which deals with the open and transparent management of personal information.

What is the new ADM transparency obligation?

Under new APP 1.7, an APP entity must include ADM information in its privacy policy where:

• the entity has arranged for a computer program to make, or do a thing that is substantially and directly related to making, a decision
• the decision could reasonably be expected to significantly affect the rights or interests of an individual, and
• personal information about the individual is used in the operation of the computer program.

If those conditions are met, APP 1.8 requires the privacy policy to set out:

• the kinds of personal information used in the operation of such computer programs
• the kinds of decisions made solely by the operation of such computer programs, and
• the kinds of decisions for which a thing that is substantially and directly related to making the decision is done by such computer programs.

Importantly, a decision may be in scope even where the computer program’s output does not replace the entire decision-making process or is subject to human review.

What is a ‘decision’ for the purposes of APP 1.7 and 1.8?

APP 1.9 provides that:

• making a decision includes refusing or failing to make a decision
• doing a thing includes refusing or failing to do a thing, and
• a decision may affect the rights or interests of an individual, whether those rights or interests are adversely or beneficially affected.

Examples of decisions that may affect the rights or interests of an individual include a decision to grant or refuse a benefit to the individual under an Act or legislative instrument, a decision that affects the individual’s rights under a contract, agreement or arrangement and a decision that affects the individual’s access to a significant service or support.

Where there is uncertainty as to whether the use of a computer program falls within the scope of the obligation, the OAIC advises APP entities to take a cautious approach and include information that complies with APP 1.8.

Not confined to AI

The obligation is not confined to AI. A computer program includes pre-programmed rule-based processes, machine learning, software, apps and word-processing tools and generative AI, including chatbots. In one OAIC example, a spreadsheet formula used to rank and triage patients for health and aged care services was in scope.

Decisions considered in scope of the transparency obligation

The OAIC’s non-exhaustive list of decisions generally considered in scope includes:

• facial recognition technology used in a retail store or stadium for watchlist matching
• recruitment software used to sort through candidate profiles and make hiring decisions
• computer programs which determine eligibility for admission to an education or training program
• reports created by artificial intelligence used to rank employee performance, determine promotions or decide bonus/remuneration outcomes
• computer programs which determine eligibility for a government benefit or housing assistance
• computer programs used to prioritise the provision of health or disability services
• differential and/or personalised pricing practices used by online retailers in the sale of significant goods
• computer programs involved in decisions about a scholarship, grant or other financial support
• programmatic advertising relating to the decision to sell significant goods or services
• computer programs used to approve or reject a loan or credit application
• computer programs involved in assessing an individual’s eligibility for insurance, and
• computer programs used to process immigration applications or enforce border controls at an airport.

How the OAIC is applying the obligation

1. When will an APP entity have ‘arranged for’ a computer program to inform or make a decision?

The OAIC considers that an APP entity may have ‘arranged for’ a computer program to inform or make a decision through:

• use of an in-house or proprietary software capability
• procurement of a third-party computer program
• configuration or customisation of existing off-the-shelf or packaged software
• authorisation of a computer program to make a determinative decision, or
• reliance on advisory outputs from a computer program.

In the OAIC’s example, an affordable housing provider uses a third-party, AI-powered software-as-a-service (SaaS) platform to process data on housing applicants, undertake predictive analytics and assist with housing decisions. Regardless of whether the provider can configure the decision parameters, it has ‘arranged for’ the platform to make decisions that significantly affect individuals’ rights or interests (access to housing), and must disclose that use in its privacy policy.

2. When is a computer program ‘substantially and directly related to making a decision’?

A computer program is 'substantially' related to making a decision where it is a key factor in facilitating the human’s decision-making, and 'directly' related where it has a direct connection with making the decision. This may be the case if the program:

• advises of an appropriate outcome of a process
• determines the outcome of a process, or
• in any other way influences the outcome of a process.

This applies whether the output is advisory or determinative. The OAIC considers that machine learning or generative AI outputs used to make significant decisions (i.e. those that significantly affect individuals’ rights or interests) will generally fall within scope unless subject to extensive human oversight and control.

Human review is not a safe harbour: the OAIC’s remuneration example

An APP entity uses a generative AI tool to assist with employee performance reviews, year-end bonuses and salary adjustments. The tool generates a profile of each sales staff member against key performance indicators and suggests a bonus and salary adjustment in a report that managers use to make remuneration decisions.

Even though HR has oversight and managers must obtain written director approval of all final decisions, the AI recommendation is relied upon to document the reasoning behind those decisions. It is therefore a key factor with a direct connection to each employee’s remuneration and is substantially and directly related to the decision.

By contrast, if managers interrogate the outputs, further review the underlying KPIs, set the tool’s parameters more narrowly, use additional evidence and document their reasons for departing from the AI recommendation, this could indicate that the output was not substantially related to the decision.

3. Who is responsible where third-party vendors are involved?

The OAIC generally expects the obligation to remain with the APP entity using the personal information to make the decision. Software providers should give clear, high-level information about how their software can be used to make decisions, so the APP entity can disclose accurately. APP entities and providers should ensure their contracts are clear about which entity retains ultimate decision-making capability. For APP entity customers relying on third party-procured AI technology, the now coined ‘AI Addendum’ needs to ensure precision, risk identification and clarity in drafting with express clauses targeted around customer data use, decision-making capabilities and model boundaries to ensure protection.

4. How should APP entities treat commercial-in-confidence information?

The obligation does not require APP entities to disclose commercial-in-confidence information about computer programs, including commercially sensitive information and trade secrets. The OAIC identifies factors that may assist, including whether the information:

• is unique to the APP entity and, if known to others, the extent to which that would detract from its intrinsic commercial value
• confers a competitive advantage on the APP entity, such as lowering costs or allowing access to markets not available to competitors, and
• is information that a genuine ‘arm’s-length’ buyer would be prepared to pay for.

In the OAIC’s lending example, how a proprietary fraud-detection program weights data points could be withheld, but the fact that personal information is used for fraud detection could not. Notably, information about an APP entity’s use of a computer program for significant decisions that would expose it to ridicule, embarrassment or public criticism does not fall within the exclusion.

Key risks

• Under-disclosure: assuming that human sign-off, or a vendor’s role, takes a decision out of scope.
• Vendor blind spots: gaps in vendor information undermine both privacy policy accuracy and breach response.
• Sensitive data: where health information or biometric templates are used, the OAIC expects this to be made clear. Health service providers again topped the list as the most frequently breached sector in the recent NDB statistics release.

Practical steps before 10 December 2026

Also on the horizon: The OAIC’s Children’s Online Privacy Code must be registered by 10 December 2026 and the Tranche 2 Reforms are moving towards Parliament. APP entities should build their ADM work with both in mind.

If you have questions on this topic, contact our team.