What 10 December means for automated decisions

All insights
Sage Insights Privacy and AIInsuranceCommunity services

What 10 December means for automated decisions

From 10 December 2026, entities subject to the Australian Privacy Principles must describe certain automated decision-making in their privacy policies. That includes significant decisions shaped by software using personal information, not just AI.

The hard part is finding where those decisions already happen.

Briana van Tilburg

Co-founder and Managing Director

Tom Arnaud

Consultant

3 minute readNovember 2026

What is the new obligation?

New rules in the Privacy Act start on 10 December 2026. They apply to Australian Government agencies covered by the Act, organisations with an annual turnover above $3 million, and some smaller ones, such as private health service providers. Most state and territory agencies are outside the Commonwealth Privacy Act. Their position depends on the state or territory law that applies.

This is known as the automated decision-making (ADM) transparency obligation. Despite the name, it can also cover decisions made by a person with substantial and direct help from a computer program.

Sometimes a computer program makes a decision about a person. Sometimes it does something that substantially and directly shapes a decision a person makes. Either way, the privacy policy must now say so. This applies when the decision could reasonably be expected to significantly affect someone's rights or interests, and their personal information is used.

The policy must describe three things.

  • The kinds of personal information used.
  • The kinds of decisions a program makes on its own.
  • The kinds of decisions a program substantially and directly helps a human decision-maker make.

It need not disclose underlying logic or trade secrets. It must give people a meaningful account of the information used and the decisions affected.

The Office of the Australian Information Commissioner (OAIC) released guidance on 30 September to help organisations prepare.

What is in scope?

“Computer program” is a broad term. The OAIC's guidance covers rule-based tools, AI and machine learning, everyday software, spreadsheets and chatbots.

Having a person sign off does not, on its own, take a tool out of scope. If its output is a key factor with a direct connection to a significant decision, it may be in scope. That holds whether the output is advice or the final answer.

The Act gives examples of significant decisions. They include granting or refusing a benefit under a law, decisions that affect rights under a contract, and decisions that affect access to a significant service or support. Vulnerability matters too. The same decision can weigh more heavily on someone who is already vulnerable.

Key issues for organisations

Writing the policy is the easy part. Finding the decisions is the work.

Most organisations do not have a list of where programs shape decisions. These tools build up over years. A triage score in one team. An eligibility rule in another. A spreadsheet someone built long ago that everyone still uses. A diagnostic imaging tool, where the smarts are just part of the functionality.

Old tools are not exempt. The rule covers decisions made from 10 December, even if the tool or its use of personal information started earlier.

Some tools sit inside software bought from a vendor. The OAIC is clear that third-party products still count. If your organisation uses one to shape a decision, your privacy policy needs to cover it.

Next steps before 10 December

  1. List the decisions that matter to people.

    Start with eligibility, priority, pricing, funding and access to services.

  2. Trace the tools behind each one.

    Ask what program feeds the decision, what personal information it uses and how much weight its output carries.

  3. Ask your vendors.

    The OAIC says vendors should give clear, high-level information on how their products can be used to make or inform decisions.

  4. Write for the people affected.

    Group similar decisions. Call out sensitive information, such as health information. Keep it readable.

Read the OAIC's guidance on automated decisions here.

What it does not do, and what comes next

This particular amendment is about transparency. It does not give people a new right to challenge a decision. It does not require organisations to tell each person directly. Other notification, complaint, access or sector rules may still apply.

But the direction is clear. The Australian Government is also developing a framework for automated decisions in Commonwealth agencies. The details are still to come.

An organisation that knows where its systems decide is ready for what comes next. It can explain a decision when someone asks. That serves the person, the organisation and the regulator at once.

If you would like a second pair of eyes on where your systems shape decisions, get in touch. We would be glad to talk it through.

General information only, not legal advice.

Sources. Privacy and Other Legislation Amendment Act 2024 (Cth) sch 1 pt 15. OAIC, APP 1 guidelines and APP 1.7 to 1.9 fact sheet, September 2026. Attorney-General, AI consumer safety priorities, 20 July 2026.

More insights

View all insights

Business continuity

What one outage means for business continuity

On 13 November, a global Microsoft 365 outage removed access to Outlook, Teams and other services for most of the day. It was an unplanned test of business continuity.

3 minute read Read article
All insights
Sage Insights Privacy and AIInsuranceCommunity services

What 10 December means for automated decisions

From 10 December 2026, entities subject to the Australian Privacy Principles must describe certain automated decision-making in their privacy policies. That includes significant decisions shaped by software using personal information, not just AI.

The hard part is finding where those decisions already happen.

Briana van Tilburg

Co-founder and Managing Director

Tom Arnaud

Consultant

3 minute readNovember 2026

What is the new obligation?

New rules in the Privacy Act start on 10 December 2026. They apply to Australian Government agencies covered by the Act, organisations with an annual turnover above $3 million, and some smaller ones, such as private health service providers. Most state and territory agencies are outside the Commonwealth Privacy Act. Their position depends on the state or territory law that applies.

This is known as the automated decision-making (ADM) transparency obligation. Despite the name, it can also cover decisions made by a person with substantial and direct help from a computer program.

Sometimes a computer program makes a decision about a person. Sometimes it does something that substantially and directly shapes a decision a person makes. Either way, the privacy policy must now say so. This applies when the decision could reasonably be expected to significantly affect someone's rights or interests, and their personal information is used.

The policy must describe three things.

  • The kinds of personal information used.
  • The kinds of decisions a program makes on its own.
  • The kinds of decisions a program substantially and directly helps a human decision-maker make.

It need not disclose underlying logic or trade secrets. It must give people a meaningful account of the information used and the decisions affected.

The Office of the Australian Information Commissioner (OAIC) released guidance on 30 September to help organisations prepare.

What is in scope?

“Computer program” is a broad term. The OAIC's guidance covers rule-based tools, AI and machine learning, everyday software, spreadsheets and chatbots.

Having a person sign off does not, on its own, take a tool out of scope. If its output is a key factor with a direct connection to a significant decision, it may be in scope. That holds whether the output is advice or the final answer.

The Act gives examples of significant decisions. They include granting or refusing a benefit under a law, decisions that affect rights under a contract, and decisions that affect access to a significant service or support. Vulnerability matters too. The same decision can weigh more heavily on someone who is already vulnerable.

Key issues for organisations

Writing the policy is the easy part. Finding the decisions is the work.

Most organisations do not have a list of where programs shape decisions. These tools build up over years. A triage score in one team. An eligibility rule in another. A spreadsheet someone built long ago that everyone still uses. A diagnostic imaging tool, where the smarts are just part of the functionality.

Old tools are not exempt. The rule covers decisions made from 10 December, even if the tool or its use of personal information started earlier.

Some tools sit inside software bought from a vendor. The OAIC is clear that third-party products still count. If your organisation uses one to shape a decision, your privacy policy needs to cover it.

Next steps before 10 December

  1. List the decisions that matter to people.

    Start with eligibility, priority, pricing, funding and access to services.

  2. Trace the tools behind each one.

    Ask what program feeds the decision, what personal information it uses and how much weight its output carries.

  3. Ask your vendors.

    The OAIC says vendors should give clear, high-level information on how their products can be used to make or inform decisions.

  4. Write for the people affected.

    Group similar decisions. Call out sensitive information, such as health information. Keep it readable.

Read the OAIC's guidance on automated decisions here.

What it does not do, and what comes next

This particular amendment is about transparency. It does not give people a new right to challenge a decision. It does not require organisations to tell each person directly. Other notification, complaint, access or sector rules may still apply.

But the direction is clear. The Australian Government is also developing a framework for automated decisions in Commonwealth agencies. The details are still to come.

An organisation that knows where its systems decide is ready for what comes next. It can explain a decision when someone asks. That serves the person, the organisation and the regulator at once.

If you would like a second pair of eyes on where your systems shape decisions, get in touch. We would be glad to talk it through.

General information only, not legal advice.

Sources. Privacy and Other Legislation Amendment Act 2024 (Cth) sch 1 pt 15. OAIC, APP 1 guidelines and APP 1.7 to 1.9 fact sheet, September 2026. Attorney-General, AI consumer safety priorities, 20 July 2026.

More insights

Business continuity

What one outage means for business continuity

On 13 November, a global Microsoft 365 outage removed access to Outlook, Teams and other services for most of the day. It was an unplanned test of business continuity.

3 minute read Read article
View all insights
Next
Next

What one outage means for business continuity