What the Public Sector AI Plan means for Australia
By Kathryn Giudes (pictured), Founder and Managing Director of ORCA Opti
The Australian Government recently unveiled the Data and Digital Government Strategy’s 2025 Implementation Plan (The Plan) – how the Australian Public Service (APS) will harness Artificial Intelligence (AI) and innovation.
Importantly, The Plan focuses on four key priority areas: AI, data, connected service delivery, and cyber trust and resilience.
But The Plan is not simply a policy update. It is a declaration of how Australia intends to use artificial intelligence as secure national infrastructure. It outlines how AI will be introduced responsibly, how public trust will be protected, and how agencies will adopt AI at scale without compromising security or governance.
The Plan is organised around three pillars. Trust. People. Tools. Together, they outline a direction that prioritises integrity above speed, and accountability above experimentation.
For those of us working in sectors like defence, critical infrastructure, healthcare and supply chain security, this plan is timely. It articulates what many of us already understand. AI becomes operationally valuable only when it is governed, traceable and designed to meet real world standards.
A Plan Built on Accountability and Evidence Instead of Hype
The most powerful part of the APS AI Plan is its commitment to responsible and sovereign use of AI. It focuses on the elements that matter most in environments where failure has consequences.
The Plan prioritises:
- transparent and evidence based outputs
- clear accountability for AI decisions
- governance models that can be audited
- secure sovereign hosting and data protection
- capability uplift across the entire APS
- improvements in procurement and shared assessments to reduce duplication
This approach recognises that AI must be safe and reliable before it can be scaled. It sets expectations that will soon inform how the Government interacts with industry, including suppliers, defence partners and SMEs across the national supply chain.
At ORCA Opti, we share the Government’s enthusiasm for exploring how generative AI can drive productivity improvements in the public service. But we also strongly support the need for measured, responsible adoption, including privacy and appropriate use of guardrails. The most powerful part of the APS AI Plan is its commitment to responsible and sovereign use of AI. It focuses on the elements that matter most in environments where failure has consequences.
In looking at sovereign AI use, each country that builds AI is doing so with a societal set of expectations. The USA is looking to reach massive commercial profit, China is looking to gain financial and military supremacy, Israel is to create military and defence capability with commercialisation a second priority, and Singapore is looking to create efficiencies and build trust within the community. Australia is about doing what is right and enforcing accountability. With sovereign AI, our societal priorities come first. Leveraging other AI, then we inherit their societal priorities.
The Plan’s priorities are aligned with our societal beliefs and help keep Australia aligned. This trial shows that benefits can be real, but only if the right supports, governance, and cultural change are in place.
We also echo the report’s call for ongoing monitoring of unintended consequences, including risk of job displacement, ethical bias, and environmental impacts, as use of generative AI expands. This approach recognises that AI must be safe and reliable before it can be scaled. It assumes visibility in the AI use and it sets expectations that will soon inform how the Government interacts with industry, including suppliers, defence partners and SMEs across the national supply chain.
Australia just made AI mandatory for the Government. Here’s why that matters for every business.
Our CTO, Phoenix Guy also welcomes The Plan. According to Phoenix, Australia just did something no other country has attempted: making artificial intelligence mandatory for every single public servant.
“The APS AI Plan 2025, launched in November 2025, requires all 220,000+ government employees to complete AI training, establishes Chief AI Officers in every agency by July 2026, and builds sovereign AI infrastructure called GovAI. As someone who’s built AI systems and watched countless “digital transformation” initiatives fizzle, this plan represents either the blueprint for how democracies adopt AI at scale, or a cautionary tale about moving too fast with too much ambition,” he says.
From philosophy to mandate
Phoenix also highlights how seven years ago, Australia released voluntary AI Ethics Principles. Fast forward to 2025, and the government isn’t asking anymore. Every public servant will be trained. Every agency will have a Chief AI Officer. Access to secure AI tools is universal. This represents a fundamental shift from “should we?” to “here’s how we will.”
“What changed? Australia’s regulatory frameworks weren’t keeping pace. The government recognised what many businesses already know: you can’t pilot your way to transformation. At some point, you have to commit. The Plan’s three pillars, Trust, People, and Tools, acknowledge that technology alone won’t drive adoption. You need governance frameworks, workforce capability, and secure infrastructure,” he adds.
“Australia’s approach is unique globally. The EU passed heavy AI regulation with penalties up to €15 million. The US has emphasised deregulation. The UK is building infrastructure but avoiding mandates. Australia is mandating government transformation while providing the tools and training to make it work. It’s not regulation for regulation’s sake, and it’s not blind faith in the market.”
The Deloitte wake, up call
Let’s talk about the elephant in the room. Just as The Plan was being finalised, Deloitte submitted a report containing AI, generated errors so egregious they could have come from a first-year student: fabricated citations, made-up court cases, and quotes attributed to judges that never existed.
According to Phenix, this incident crystallised the core tension: How do you accelerate adoption while maintaining quality? The Plan’s answer is mandatory transparency. All contractors must now disclose AI use upfront, and suppliers remain fully responsible for deliverables regardless of whether AI was involved.
“For Australian businesses, this is the crucial lesson: transparency is becoming a competitive advantage, not a liability. The firms that thrive won’t be those that use AI most extensively, but those that use it most responsibly and can prove it,” he explains.
What this means for business
We believe the opportunities are substantial. The Plan creates immediate demand for compliance platforms, training providers, cybersecurity firms, and systems integrators. Companies like ORCA Opti, which already provides AI governance tools for defence and financial services, are well positioned with technology that monitors for compliance breaches and provides audit, ready documentation.
The emphasis on Australian infrastructure creates natural advantages for local tech companies. But there are legitimate concerns: job displacement fears and tight implementation timelines that could lead to box-checking rather than genuine transformation.
The real test ahead
The APS AI Plan represents Australia’s bet that the Government can be an exemplar rather than a laggard in AI adoption. If it works, if those 220,000 public servants become genuinely AI-capable while maintaining public trust through transparency and accountability, it could become a model for other mid-sized democracies navigating similar challenges.
If it fails, if mandatory training becomes perfunctory certification, if Chief AI Officers become bureaucratic overhead, if the speed of implementation compromises safety, it will validate the critics who argue you can’t mandate innovation.
For business leaders, the message is clear: the era of voluntary AI governance is ending
What the Government is doing with mandatory transparency, impact assessments, and accountability frameworks will eventually extend to regulated industries and high-risk private sector applications. The companies that get ahead of this shift, that build robust AI governance not because it’s required but because it’s good business, will have a significant advantage.
Australia isn’t just launching an AI plan. It’s running an experiment in democratic AI adoption at scale, with real deadlines, real accountability, and real stakes. The rest of the world is watching. Australian businesses should be too.
