Information Awareness Week Canberra Seminar – Opening address

Read Freedom of Information Commissioner Alice Linacre’s prepared speech to the Information Awareness (IAW) Canberra Seminar on 18 March 2026.

I’d like to start by acknowledging the traditional owners of the land we are meeting on today, the Ngunnawal people, I’d like to also acknowledge that there are other families which have strong and abiding connections with this country. I’d like to pay my respects to Elders past, present and emerging and to any First Nations people here today.

Freedom of Information is a statutory right to access documents held by government.

That right is premised on records being properly held and being able to be found, I am absolutely thrilled to be speaking at today’s Information Awareness Week Canberra Seminar. Thank you to RIMPA for the invitation and to the National Archives of Australia for hosting us in this beautiful building.

Todays’ theme, that information quality, governance, and records management form the foundation of successful artificial intelligence (AI) outcomes, is a timely conversation to have. Governance of AI is coming into focus as we seek to use it to innovate and build efficiencies.

In the Digital Transformation Agency’s (DTA) guidance from just last week, issued to support AI project success, Lucy Poole, Deputy CEO of the DTA is quoted as saying:

“By considering governance from the outset, how it will fit with existing systems, and what success looks like, you can turn what works into something that lasts and delivers real benefits for government and the people it serves.”

This focus on governance is entirely appropriate and relevant to records management, and to the conversation about how will AI benefit from the existing discipline in this space.

AI requires information and records to learn from and be trained. Good record-keeping will underpin the ability for AI to succeed. AI will produce records including documents. The importance of those records and documents will continue to grow as the capability of AI increases and its uses expand. Underneath all of this sit principles of democracy that require accountability and transparency.

Those principles will be more important than ever as the pace of decision-making grows and the need and desire from the public to understand decision-making increases.

This is particularly so as decision-making speed and volume increases, the ability to understand and identify errors in those decisions becomes more critical, an example of this in automated decision-making (ADM) we saw through Robodebt.

Today, I want to talk to you about 3 themes:

1. Agencies, in the face of emerging technology, should wrestle with and remain committed to good accessible record-keeping.
2. Access to documents is important to provide understanding of AI and ADM in decision-making.
3. Proactive release of information has never been more important. Publicly released information provides credible information for AI to synthesise, and it provides a ‘source of truth’, for people to check AI outputs against.

Good Accessible Record-keeping

In 1979, a Senate committee report outlined three reasons why FOI is important and they remain relevant today:

1. FOI allows individuals to see what information government holds about them, and to seek correction of that information if they consider it wrong or misleading.
2. FOI enhances the transparency of policy making, administrative decision making and government service delivery; and finally
3. A community that is better informed can participate more effectively in the nation’s democratic processes.

The Freedom of Information Act 1982 enshrines the right of access and the obligation to proactively publish certain information playing an important role in facilitating transparent government decision making.

The Objects of the FOI Act include requiring agencies to publish information; providing a right of access to documents; and increasing scrutiny, discussion, comment and review of the government’s activities. These objects are essential to building and maintaining the openness, responsiveness and integrity of government agencies, which the OECD has found to be key drivers of trust in public institutions.

These objects are also essential to our system of responsive, representative democracy.

So, the FOI Act is a critical part of our democratic toolbox, but the mere existence of the framework is not enough. Administration of that framework has a critical dependency in good record-keeping.

As AI use expands in government, the need to have good record-keeping which allows identification of the right documents, which set out reasons and explanations for decisions and policy, is essential.

The desire to know more and find out what information government holds is reflected in the 43,456 FOI access requests last year – a 25% increase on the previous year and the increase in FOI decisions made in 2024-25 – 25,211 decisions (18% more than the preceding year).

For a person to understand and engage in government decisions, access to these types of documents has always mattered, but when the decision is either automated or involves input from AI, accountability and transparency are never more vital.

As experts in this field, you will think about with what records should be retained, I would ask in that conversation you remember that people’s ability to understand our government will be more important as government considers and implements AI into the public service.

Ultimately, FOI is a human-centred administrative law, it sits at the interface between government and people. The more documents that are accessible, that evidence the workings of government (and where that is about use of AI) the better.

The approach to these novel record-keeping, FOI and Privacy obligations in the joint OAIC/NAA messaging apps report released in March 2025, engaged with these issues as they pertained to that emerging technology. AI will pose another such challenge.

While AI may present challenges to record-keeping practices, including when it comes to volume, speed and formality of record-keeping, AI may also represent opportunities in searching for and retrieving records.

In FOI practice we see disputes in the OAIC’s review jurisdiction, where agencies have conducted searches following receipt of an access request, but either to continue to search would be an unreasonable diversion of resources, or documents cannot be found. Both of these outcomes can be extremely frustrating for an FOI applicant, but also for an agency attempting to process a request. Our guidelines make clear that the steps that are reasonable will be contextual, but equally poor record-keeping is not a basis to frustrate FOI access request attempts.

In reality, where those barriers to searching exist (bad naming practices, storage on local drives etc) there is not much that can be done to find the documents. A future for FOI could be the breaking down of walls between record-keeping frameworks and broadly more searchable environments, using AI.

Agencies, in the face of emerging technology, should wrestle with and remain committed to good accessible record-keeping, this pressure will increase at the same rate as the data or information-holdings. Ultimately, it is important to remember that AI is a tool and that the principles of transparency and accountability and our legislative frameworks that instil those, remain primary and technology neutral in the objects they legislate to instil.

I am hopeful that AI may produce an amazing searches solution, but of course that will only be as good as the information and documents available to search.

AI and ADM decision-making

It is useful to be clear about what I refer to as AI as opposed to ADM.

As a general term, AI refers to the ability of machines to perform tasks which normally require human intelligence.

Generative AI is a category of AI that can create new content such as text, images, videos and music. It gained global attention in 2022 with text-to-image generators and large language models (LLMs). These technologies are fast paced in their development and impact.

However, AI in the form of ADM has been around for a long time.

By ADM, I mean the use of technology to automate decision-making processes. It is used across government in areas such as social services, taxation, aged care and veterans’ entitlements. ADM predates modern AI and can range from simple rules-based systems, e.g. calculating payments using legislated formulas, to more advanced systems using machine learning to make predictions or support decisions.

ADM can increase the efficiency and consistency of decisions across the public and private sectors.

While ADM doesn’t always involve AI, both ADM and AI raise similar policy concerns, including risks relating to bias, discrimination, errors, transparency, and accountability.

The OAIC recently published a report detailing how transparent 23 Australian Government agencies are about their use of ADM in decision-making processes.

This involved reviewing their websites, AI agency transparency statements and Information Publication Scheme (IPS) agency plans.

We found that all reviewed agencies published IPS information on their website.

However, inclusion of ADM information through proactive release under the IPS and on their agency websites was limited:

• 100% (23) of agencies complied with the requirement to publish IPS information
• 17% (4) of agencies authorised to use ADM, disclosed that they use ADM in a decision-making process in their IPS information
• 39% (9) of agencies (in addition to the 4 above) made reference to ADM in their IPS information but did not confirm that they used it
• 43% (10) of agencies did not mention ADM in their IPS information
• The agencies that disclosed in their IPS information that they used ADM were not clear about how they used it
• Finally, 0% (0) We did not find any agency that had any published guidelines or policies on ADM.

We outlined recommendations to help agencies publish policies and procedures under the FOI Act proactively, about their ADM use; to foster transparency and open government:

• Recommendation 1: All agencies authorised under legislative statutes to use ADM should publish this information as part of the IPS. This should include the statute that grants them this power, and whether or not they utilise ADM to provide information and services to the public.
• Recommendation 2: As part of their IPS, agencies should clearly state the types of ADM they use to make automated decisions, not just AI (i.e. from simple calculators to machine learning).
• Recommendation 3: Agencies that use ADM should publish, as part of their IPS, both a list of decisions they use ADM for, and relevant and easy to understand examples, so the public can better understand how it is being utilised to make decisions that affect them.
• Finally, Recommendation 4: Agencies that use ADM should publish, as part of their IPS, any policies that clearly set out the principles for when and how they use ADM to make decisions and/or recommendations affecting members of the public.

Transparency around decision-making is ultimately a matter of trust. The OECD’s trust survey shows 46% of Australians have a high or moderate degree of trust in the federal government. Trust is lower among those experiencing disadvantage, such as financial distress, and those who have lower education levels.

There is a long way to go to build trust in AI. Although a bit dated now, our 2023 Australian Community Attitudes to Privacy Survey found:

• Australians are cautious about the use of AI to make decisions that might affect them, with 96% saying there should be some conditions in place first.
• Only 1 in 5 were comfortable with government agencies using AI to make decisions about them.

In research released in 2025, KPMG found that Australians are some of the least trusting of AI, so while half of Australians identify using AI, of those only 36% trust AI and 78% of people are concerned about negative outcomes.

In the 2025 Australian Information Commissioners and the ACT Ombudsman cross-jurisdictional study of community attitudes on access to government information: in the space of automated decision-making and AI, the majority (over 75%) across all jurisdictions agree that the government must publicly report on any technology used to inform decision-making, with the highest agreement for the Commonwealth Government.

Evidence-based decision making by government is a strong driver of trust.

Where AI is involved in government decision-making, the opacity of these systems can pose significant challenges for agencies seeking to provide a meaningful explanation about how decisions are made.

We must tell individuals that AI is in use, how it’s being used and ensure that when they exercise their right to obtain information under a FOI regime, that the information exists and can be requested by them.

Proactive release of information

As AI becomes more ubiquitous the level of trust and reliance will be aided by the capacity for people to have access to source material, to be able to verify and inform themselves sufficiently to identify problems with the AI interpretation or potential hallucinations. Proactive release of information may assist in that verification and confidence.

Proactive release of information may also ensure AI has accurate information upon which to build its capability.

In June 2024, the Information Publication Scheme 2023 Survey was published – with a 94% response rate.

The Information Publication Scheme (IPS) requires agencies to publish a broad range of information on their website and to proactively publish other information. The IPS continues to be an important element in ensuring information held by Australian Government agencies is managed for public purposes and is treated as a national resource.

Overall, as noted by the Australian Information Commissioner the survey showed that there is a strong commitment across the APS to the IPS and to proactive disclosure.  One of the best takeaways was that 94% of agencies have reviewed the operation of the IPS in their agency, which is an increase from 82% in 2018. However, the results also told us that only 29% of agencies have adopted a strategy for increasing open access to information they hold, down from 35% in 2018.

At the time the Information Commissioner said:

“These results indicate that the intent of the scheme is not fully realised.  Collectively there is work to do to ensure that the objects of the FOI Act are realised. The FOI Act aims to increase public participation in government processes and increase scrutiny, discussion, comment and review of government activities. We have a tool to promote open government in the FOI Act and collectively we can achieve that ambition.”

Proactive publication is something that the OAIC has expressed many views on and is aligned with our current regulatory priorities. Which include ensuring timely access to government information and strengthening the information governance of the APS, as well as rebalancing power and information asymmetries and rights preservation in new and emerging technologies.

In 2021 Information Commissioners and Ombudsmen across Australia released principles for proactive release of government held information that acknowledged that to the greatest extent possible information should be published promptly and proactively at the lowest reasonable cost without the need for a formal access request.

The benefits being that proactive disclosure informs the community, increases participation and enhances decision-making, it builds trust and confidence, improves services delivery and promotes efficiency by reducing the administrative burden on departments and agencies and the need for people to make formal information access requests.