AI is increasingly becoming part of everyday work across government.
For Dr Ian Oppermann, Commissioner, Australian Competition & Consumer Commission, that means AI literacy can no longer be treated as a specialist or optional capability. Public sector leaders need frameworks that help people understand where AI can be used, which controls are appropriate and who remains accountable for the outcome.
Ahead of Digital Leadership Day Federal 2026, Public Sector Network spoke with Ian about the accelerating pace of technological change, the practical meaning of AI literacy, how agencies can build useful guardrails without slowing innovation, and why different AI use cases require different levels of control.
Ian will continue this discussion in the panel “AI Literacy: Leveraging Capability for Responsible Use Across Government”, joining Alice Linacre, Freedom of Information Commissioner, Office of the Australian Information Commissioner, Andrew Watson, Deputy Commissioner and Chief Data Officer, Australian Taxation Office, and Monita Lal, Head of AI Delivery and Enablement, Department of Finance.
The session will explore why AI literacy is becoming a core public sector capability and how agencies can build workforce confidence while maintaining governance, security and public trust.
Explore the event: Digital Leadership Day overview | View the agenda
Q&A with Dr Ian Oppermann
Public Sector Network: You’re speaking at Digital Leadership Day. What made this panel worth your time, and what kind of conversation are you hoping to have in the room?
Dr Ian Oppermann:
We certainly live in an age where digital is part of every conversation, and it’s been a truism for some time, but it’s even more true now that the world is changing faster and faster every single day.
The rate of change is actually accelerating, and so the question is: how can public servants and leaders in public service cope with that change, particularly when so many things are changing in so many different ways?
The conversation I’m hoping to have is really about that skill set and framework for thinking our way into a constantly evolving world where change is actually accelerating.
Public Sector Network: AI is front and centre in that change. What does being AI literate as a public servant actually look like, and where is the biggest gap today?
Dr Ian Oppermann:
I think you get a very different response to that question today than you would have a year ago, or even two or three years ago.
It used to be that AI was something you would deliberately choose to engage with. You would experiment with AI to try to improve a service, do things differently or identify some insights.
Now AI is happening all around us.
Every time you open a browser, your browser has been updated with a new element of AI. Every time you use a piece of software, it has increasingly been updated to include AI.
So it’s really unavoidable.
We’re seeing it not just in the tools we use, but in the way we interact, the way we engage with community, the way we engage with society and the way we engage with companies.
It’s happening to us, it’s happening around us and it’s happening really everywhere. There is no part of society, economy or community which is not being impacted by AI.
So the question is: how do you appropriately engage with AI, and what does literacy mean?
Today’s answer is something along the lines of: I’m aware of it, I experiment with it, I have guardrails, and I have a way of thinking about interacting with it and using it.
But going beyond that, I’ve got systems and controls in place which allow different sorts of uses of AI in different environments, with different consequences.
It is more than just creating a sandbox and experimenting, which is sort of yesterday’s approach.
You might have people using the AI equivalent of butter knives, people using the AI equivalent of steak knives and people using the AI equivalent of samurai swords, with different levels of skill, domain knowledge, expertise, legal frameworks and care around the products that are created from AI.
It’s really about creating a framework within the risk appetite and strategy of the organisation that allows people to use appropriate tools for appropriate purposes, with appropriate controls, while still letting people use those more powerful tools if they’re equipped for it.
Public Sector Network: As access to generative AI expands across the public service, which guardrails genuinely reduce risk and which simply slow people down without making decisions any safer?
Dr Ian Oppermann:
The approach I always take is thinking about everything around the data before you use it.
Where did it come from? How did it get to you? What controls need to be on that data?
That data could be rows and columns, a report or anything with a data element that could be used by AI.
You need to think about how fit for purpose it is and whether it is appropriate for what you want to use it for.
Then, when you create a data product, it could be a report, a recording of a meeting, an insight, an action or a decision. You need to think about the consequences of using that product.
If the consequences are very significant or irreversible, you have to put some really strong controls around them.
If the consequences are relatively benign or readily reversible, then the controls need not be quite so significant.
It’s really about having frameworks of controls that are appropriate guardrails and allow people to do more powerful things with AI.
Those frameworks of control sit at the intersection of domain expertise, expertise around use of the tool, an understanding of governance and the legal authorising framework, and a strong understanding of the consequences, how to mitigate them and, in some cases, how to reverse them.
It sounds complex, but it’s actually a simple set of guardrails that creates frameworks where you can go really slowly and experimentally, or really fast and do really powerful work.
Public Sector Network: Who is accountable when an AI-assisted decision goes wrong, and how do you build accountability across the whole value chain?
Dr Ian Oppermann:
That used to be a really easy question to answer.
If I am someone using an AI-driven tool, I’m responsible. I’m always responsible for the outcome or the output and the consequences of the use of that output or data product.
It gets a little trickier when we start to talk about more than a single interaction between a human and an algorithm.
If this algorithm interacts with that algorithm, which interacts with another algorithm, you can get into some really interesting thought experiments.
What happens when one algorithm starts interacting with the original algorithm, and then I stop participating as a user? I move to another department, or I’m no longer there. Who is responsible?
That is at the extreme end of thinking about where responsibility lies.
But somewhere between that simple interaction and those really complex thought experiments, there is still this concept that whoever initiated it, provided authority and chose the control framework is responsible.
Are they the only person responsible? That’s a more interesting question.
Within the level of control and understanding I’ve got, if I reasonably expect certain activities to take place, then I can take reasonable actions. I’m still responsible.
If something happens that really is out of my control or I really don’t understand, then we shouldn’t be doing it. If we do, there are other questions about whether the guardrails were appropriately established, whether the AI product was safe to use and whether the data product was safe to use.
Right at the moment, we’re still at: “I use it, I’m responsible. I’m responsible for the output and therefore the outcome.”
But I think we really need to be alive to the fact that this will change over time.
Public Sector Network: How do agencies balance innovation and speed with governance and security?
Dr Ian Oppermann:
Again, it comes back to understanding how many frameworks or controls you want to have.
There could be low, medium and high. There could be none, low, medium, high and very high. You can slice it however makes sense for your organisation.
But those frameworks of controls need to be defined by people with technical, domain and tool competence, as well as an understanding of governance.
There is also an obligation to think through consequences and mitigations, all the way through to lighter use cases or even a no-control environment where you’re using public data and public AI tools.
It’s important for those frameworks to exist.
It’s a bit like giving someone who drives a car brakes, seat belts and headlights. Those things allow you to drive faster because you can control what is happening.
You control the use of the output.
By having gradations between a no-control environment and a very high-control environment, you can actually say, “This is how fast, how precise and how sharp the tools are that you’re allowed to use.”
That really allows things to go faster.
If we only have an on or off switch, or if we only have one speed, then it really holds organisations back.
People become worried about the consequences of what they do, so they either don’t use the tools or they do it through shadow uses of AI.
It’s really important to agree those frameworks of control, drag them out into the light and make sure everybody knows what they are.
AI Literacy: Leveraging Capability for Responsible Use Across Government
Ian’s comments go directly to the challenge at the centre of his upcoming Digital Leadership Day Federal panel.
As generative AI becomes embedded in everyday workflows, AI literacy is increasingly about more than knowing how to use a tool.
It means understanding the data behind it, the consequences of the output, the level of risk involved, which controls are appropriate and who remains accountable.
The panel “AI Literacy: Leveraging Capability for Responsible Use Across Government” will explore:
- why AI literacy is becoming a core public sector capability, not just a technical skill
- enabling safe and guided adoption of generative AI at scale
- accountability for AI reliability across the innovation value chain
- building workforce confidence through practical experience
- balancing innovation with governance, security and public trust
Ian will be joined by Alice Linacre, Andrew Watson and Monita Lal, with Jennifer Mulveny, Director of Government Relations, Asia-Pacific, Adobe, moderating the discussion.
Leadership in an AI-enabled public service
The AI Literacy panel forms part of the broader Digital Leadership Day Federal 2026 program, which brings together senior public sector leaders to examine how leadership itself needs to change as technology, AI, workforce expectations and operating environments evolve.
For leaders, the challenge is increasingly not whether AI will affect their organisations. As Ian notes, AI is already becoming embedded in the tools and systems people use every day.
The leadership question is how to build the confidence, governance and practical capability that allow teams to use those technologies responsibly.
Digital Leadership Day Federal takes place on Wednesday, 21 October 2026 in Canberra and is designed for senior public sector leaders, including directors and executives.
Explore the event: Digital Leadership Day overview | View the agenda
Help your peers
Share what you've learned with fellow public servants