top of page
Search

Snapshot: Australia’s National AI Plan (Dec 2025)

  • Gov+AI
  • 3 days ago
  • 4 min read

The Australian Government’s National AI Plan, released on 2 December 2025, sets a decade-long framework to embed artificial intelligence across the economy. The Plan was needed and is a welcome step. However it falls short on translating aspiration to rigorous implementation.



The Headline:

Investment: With an additional $29.9 million investment (on top of $460 million in prior funding) it aims to grow Australia's GDP through the use of AI in infrastructure, skills, and small business adoption.


Oversight: Rather than new laws, it leans on existing regulations like the Privacy Act 1988, voluntary ethics principles, and oversight from a new AI Safety Institute. This is a good step forward, but the lack of overarching legislation and “opt-in” and self-reporting elements is a weakness.


Key Elements of the Plan

Structured around three pillars—capturing opportunities, spreading benefits, and ensuring safety—the plan details specific allocations (each will get a deeper dive in future blogs):

  • Capturing opportunities: $20 million funds the AI Safety Institute for system testing and benchmarking. It supports GovAI cloud services and the National AI Centre, alongside data centre guidelines to bridge the gap from 100MW current capacity to 20GW needed by 2030.

  • Spreading benefits: $58 million targets 300,000 new AI-skilled workers via the AI Skills Academy and AI Adopt Lab for SMEs, offering free tools and training.

  • Ensuring safety: An AI Expert Advisory Group promotes voluntary AI Ethics Principles, with the Institute scanning for risks like algorithmic bias. No high-risk categorisation or prohibitions.


How might this translate in the real world?

  • Health: AI pilots on My Health Record data aim to predict patient risks, with trials showing potential 10-15% reductions in admissions.

  • Agriculture: $15 million backs precision tools like Queensland drone monitoring, yielding up to 20% crop improvements in trials.

  • Education: Scaled AI learning apps from COVID pilots target 1 million students yearly for personalised support.


Enforcement and Compliance

Enforcement capacity is a core weakness. The voluntary framework lacks binding mechanisms—no fines, audits, or compliance thresholds beyond self-reporting to the Safety Institute. While principles echo global standards (e.g., OECD AI guidelines), absence of mandated reporting means uptake depends on goodwill.


The ABC reports what could have been...

In September last year, former industry minister Ed Husic flagged 10 "mandatory guardrails" under development, which would include requirements for high-risk AI developers to create risk-management plans, test systems before and after deployment, establish complaints mechanisms, share data after adverse incidents, and open records to assessment by third parties. Those guardrails were intended to operate under a standalone AI act that could be used to categorise technologies based on risk, with strict rules over high-risk AI and less regulation to encourage lower-risk tools. But the government has stepped back from that path, committing in the National AI Plan to instead using "strong existing, largely technology-neutral legal frameworks" and "regulators' existing expertise" to manage artificial intelligence in the short term.


As a result there will be no enforcable, dedicated legislative or regulatory oversight. Instead, Government intends to leverage existing technology-neutral laws while introducing targeted reforms. Two Birds explain that agencies and regulators will retain responsibility for identifying, assessing, and addressing potential AI-related harms within their respective policy and regulatory domains. Central to this is the AI Safety Institute (AISI), which will assess upstream risks (capabilities, datasets, system design) and downstream harms, support specialist regulators, and coordinate major incident responses. It is likely to become a practical reference point for "what good looks like" in AI testing and documentation. 


Beyond AISI, the Plan highlights the regulatory activity already in motion, including:

  • Privacy law reform, with direct implications for profiling, automated decision-making and data governance.

  • Consumer protection and product safety, confirming AI systems fall within existing guarantees and safety rules.

  • Online safety, including proposed enforcement tools for deepfakes, nudify apps and AI-facilitated harms.

  • Copyright, where the Government has ruled out a broad text-and-data-mining exception.

  • Sectoral regimes, including health, medical devices, finance and critical infrastructure.

  • The overall message is that AI is already regulated. Future reforms will be aimed at applying fit-for-purpose legislation, strengthening oversight and addressing national security, privacy and copyright concerns.


However, as discussed in earlier Gov+AI Blog Posts, overarching AI legislation has its place too, and that is something Australia is yet to adopt.


Summary


This is a welcome Plan. However while compliance requirements are minimal: businesses will adopt ethics codes at their discretion, with the Institute providing non-binding alerts. This means it is likely to be received as light-touch regulation which suits innovation but falters on enforceability, hinging on future voluntary buy-in.


All in all, it is was a necessary first step, but if the future is AI, then stricter rules and oversight will be essential. Dropping the original guardrails approach makes this just more difficult.

 

Sources

[2] National AI Plan | Department of Industry Science and Resources https://www.industry.gov.au/publications/national-ai-plan

[4] AI Plan for the Australian Public Service 2025: At a glance https://www.digital.gov.au/policy/ai/australian-public-service-ai-plan-2025/at-a-glance

[5] Artificial intelligence to be managed through existing laws under National AI Plan https://www.abc.net.au/news/2025-12-02/national-artificial-intelligence-plan-growth-existing-laws/106086474

[7] Australia has a National AI Plan. Now What? https://www.acc.com/australia-has-national-ai-plan-now-what

[9] Australia's National AI Plan: big ambitions, but light on details https://www.whitecase.com/insight-alert/australias-national-ai-plan-big-ambitions-light-details

[10] Australia's national AI plan has just been released. Who exactly will benefit? https://theconversation.com/australias-national-ai-plan-has-just-been-released-who-exactly-will-benefit-270981

 
 
bottom of page