Skip to content

Australian Government AI Resources (Federal, State & Territory)

Purpose: Directory of official federal, state and territory AI policies, strategies and assurance frameworks Audience: Public sector organisations, cross-jurisdictional projects and government vendors | Time: 30-45 minutes

This page curates official Australian government resources on AI for public sector use, covering federal agencies and all states and territories. It focuses on whole-of-government strategies and policies, AI assurance frameworks, information/privacy guidance from statutory bodies (e.g., information commissioners, ombudsmen) and records management directives. Where education-specific positions exist, they're included because they are often the most mature sector guidance.

Last verified: February 2026. Links point only to official government or statutory sources.


General Guidance

Federal, state and territory resources are designed to work within each jurisdiction's legislative and policy settings. If you operate across multiple jurisdictions—or consume services from another state—start with the local requirements and then map to national artefacts (see Alignment below). Treat AI projects as socio-technical: combine policy, risk, privacy, security, procurement, records management and assurance practices from the outset.

Note: Policies and guidance change rapidly (especially around generative AI). Always check for the latest version and any departmental circulars or implementation notes before relying on a specific document.


How to Use This Page

  1. Identify your jurisdiction and applicable agency cluster (e.g., central DPC/DTF/Finance, Education, Records Authority, or Information Commissioner).
  2. For each item:
    • Confirm status/version and whether it is mandatory (policy/standard) or advisory (guidance/toolkit).
    • Perform or update risk and privacy impact assessments (PIA/DPIA) and recordkeeping actions.
    • Where AI supports decisions that affect rights or benefits, apply administrative law duties and human-in-the-loop controls.
  3. If your work spans states, use the Alignment guidance to harmonise.

Federal (Australian Government)

For a detailed overview of Australian AI legislation, see our AI & Australian Legislation page.


New South Wales (NSW)

Australia-first AI workplace legislation: passed 12 February 2026; primary duties not yet commenced

The Work Health and Safety Amendment (Digital Work Systems) Act 2026 (assent received 18 February 2026) makes NSW the first Australian jurisdiction to impose express WHS duties on employers using AI, algorithms, automation and digital work platforms. As of June 2026, the primary duties have not yet commenced. Commencement requires SafeWork NSW to complete public consultation on guidelines for WHS entry permit holders, publish the guidelines, and observe a statutory waiting period of at least one month. Those guidelines are still under development. Monitor SafeWork NSW and legislation.nsw.gov.au for proclamation announcements. Two primary duties are established:

1. Primary Duty (s 19(3)(c1)): A person conducting a business or undertaking (PCBU) must ensure, so far as is reasonably practicable, that the health and safety of workers is not put at risk from the use of digital work systems by the business.

2. Work Allocation Duty (s 21A): PCBUs using digital work systems to allocate work must ensure, so far as is reasonably practicable, that the allocation does not put workers' health and safety at risk. PCBUs must specifically consider whether digital work systems result in:

  • Excessive or unreasonable workloads
  • Use of excessive or unreasonable metrics to assess and track worker performance
  • Excessive or unreasonable monitoring or surveillance of workers
  • Unlawful discriminatory practices or decision-making

Union inspection rights: WHS entry permit holders can require reasonable assistance to access and inspect digital work systems where a safety breach is suspected (48 hours' notice required).

"Digital work system" is broadly defined as an algorithm, AI, automation or online platform used to allocate work. Applies to all NSW employers using AI scheduling, algorithmic management, digital performance monitoring or automated work allocation.

Employers should map their AI-driven work allocation systems now in preparation for commencement. (Sources: Norton Rose Fulbright, Hamilton Locke, NSW Parliament)


Victoria (VIC)

Victoria — $3.5M AI and Deeptech Pre-Accelerator Initiative (announced 17 June 2026)

The Victorian Government announced $3.5 million for nine pre-accelerator program operators supporting AI and deeptech startups at the Victorian Startup Gala on 17 June 2026. The announced funding is for operators rather than direct startup grants, with each provider eligible for up to $400,000. The operators are Boab AI, Boson Ventures, Cicada Innovations, CoLabs Australia, HEX, Illume Ventures, Jumpstart Studio, MedTech Actuator and RMIT DiscoveryHUB.

A featured program is VICTOR:AI — an eight-week cohort for AI-native startups offering access to AI tools, co-working space and milestone-based grants. This initiative is consistent with Victoria's AI Mission Statement (announced 30 January 2026), positioning Victoria as a national AI leader.

Provider intake arrangements and timing vary. The announcement did not specify which programs were accepting applications or confirm that funding had been disbursed. See djsir.vic.gov.au for the provider list and announced program details (accessed 21 June 2026).

Victoria — workplace surveillance and AI proposal (announced 20 July 2026; pre-legislative)

The Victorian Government announced that Labor would introduce workplace surveillance protections covering AI-enabled monitoring and automated work decisions. This is an announcement of proposed laws, not a bill. A search of the Victorian legislation register's bills list for 2026 on 8 September 2026 returned no workplace surveillance bill, so nothing in this entry creates a current obligation.

The announcement describes proposed protections including:

  • A worker's right to be told when they are being monitored, and employer consultation before surveillance is introduced or changed
  • Surveillance at work only for a legitimate purpose, with covert surveillance confined to suspected unlawful activity and strictly controlled
  • Biometric data collected only where no less intrusive option exists, and not used to read emotions unless there is a legitimate reason such as fatigue monitoring for truck drivers
  • No use of AI to track matters such as bathroom breaks, a limp or pregnancy without a specific, legitimate purpose
  • Stronger protection against workplace surveillance and AI being used to discriminate by race, gender, sexuality, religion or other personal characteristics
  • Human review of any significant automated decision made using surveillance data that affects a worker's rights, and worker access to their own surveillance data

Organisations using AI scheduling, biometric monitoring or algorithmic management in Victoria should watch for a bill and its final scope. See the official Victorian announcement (accessed 8 September 2026) and the Victorian Government workplace surveillance page.

Victoria — online platform accountability proposal (announced 19 July 2026; pre-legislative)

The Victorian Government announced two proposed reforms directed at social media and AI platforms. A search of the Victorian legislation register's bills list for 2026 on 8 September 2026 returned no bill for either reform, so neither is current law.

  • VCAT "demasking" orders: the Victorian Civil and Administrative Tribunal would be given power to make orders requiring social media companies to reveal the identity of anonymous users accused of online vilification.
  • Lower threshold for claims by minors: the current requirement to prove permanent impairment of 10 per cent or more would be removed for negligence claims brought on behalf of minors against social media and AI providers. Whether the change extends beyond children to adults is stated as a matter to be considered as the law is drafted.

The announcement says both reforms will be developed following targeted consultation with VCAT, the courts and other stakeholders. See the official Victorian announcement (accessed 8 September 2026).

Victoria — premier change (28 July 2026): open question for pending AI-adjacent bills

Ben Carroll was sworn in as Victoria's Premier on 28 July 2026, replacing Jacinta Allan, with Gabrielle Williams as Deputy Premier. As at 8 October 2026, no statement has been found on whether the Carroll Government will keep the two AI-adjacent proposals above (workplace surveillance and AI-in-hiring limits; online safety for minors). Both remain pre-legislative ahead of the November 2026 state election. Do not assume either will go ahead or be dropped until a bill is introduced or the Government states a position.


Queensland (QLD)

Queensland — Manufacturing AI Uplift Trial ($750,000, announced 30 September 2026)

The Queensland Government is investing \(750,000 in a Manufacturing AI Uplift Trial Program, delivered by the ARM Hub, to help Queensland manufacturers implement practical AI solutions. Expressions of interest close 30 October 2026. This is **separate from** the still-unnamed ~\)10 million small and family business AI adoption program announced in June 2026, which remains announced but not yet open for applications. See AI Grants & Funding for detail.


South Australia (SA)

South Australia — Data Centre and AI Infrastructure Act proposed (June 2026)

The South Australian Government's June 2026 Data Centre Strategy commits to consulting on a proposed Act covering development approvals, energy, water, security and local benefits. The Act remains proposed; existing planning mechanisms and the proposed future legislation have different legal status.

See AI Data Centres and Compute Governance in Australia for a detailed, source-led summary and the official South Australian strategy.

South Australia — Royal Commission into Artificial Intelligence commenced (1 October 2026)

The South Australian Government's Royal Commission into Artificial Intelligence — reported as the first of its kind in Australia — commenced on schedule on 1 October 2026, with a final report due no later than 1 July 2027. Three commissioners have been named: Dr Iain Ross (chair; chair of the federal Net Zero Economy Authority, former president of the Fair Work Commission, and former Federal Court and Victorian Supreme Court judge), Kate Pounder (former Tech Council of Australia CEO), and Professor Christopher Manning (co-founder of Stanford's Human-Centered AI institute). The announcement said the Commission would focus on policy and regulatory settings at state and national level, education including schools and higher education, public services provided by the state including health, skills and workforce, and AI-related infrastructure including energy transformation and water usage.

A royal commission is an inquiry. It does not itself create obligations for organisations using AI.

See the official Department of the Premier and Cabinet announcement (dated 11 August 2026) and reporting on the commissioner appointments: ABC News, Capital Brief and InDaily (all accessed 4 October 2026, via search).


Western Australia (WA)


Tasmania (TAS)


Australian Capital Territory (ACT)


Northern Territory (NT)


Alignment Across Jurisdictions

Many jurisdictions explicitly reference or align to the National Framework for the Assurance of AI in Government (endorsed by Data & Digital Ministers on 21 June 2024). When planning cross-jurisdictional work:

  • ✅ Use the local state/territory policy and assurance artefacts to determine mandatory steps.
  • ✅ Map to the national framework to ensure consistent governance, documentation and transparency across borders.
  • ✅ Reconcile differences in privacy law, administrative law practice, procurement rules and records obligations by documenting equivalences and any additional local controls.
  • ✅ For agencies consuming services or models hosted in another jurisdiction, require suppliers to provide evidence of AI assurance against both the provider's and customer's frameworks, plus PIA/DPIA, security risk assessment and records/disposal mapping and a clear ADM accountability model (human oversight, contestability, redress).

Maintenance & Contributions

  • Keep this page updated with the latest versions and add sector-specific positions (health, justice, education) when central agencies or statutory bodies publish them.
  • Prefer official sources (government and statutory bodies). Include the document title, issuing body, version/date and whether it is mandatory or guidance.

Disclaimer & Licence

Disclaimer: This directory provides links to official Australian federal, state and territory government AI resources. SafeAI-Aus has exercised care in curation but does not guarantee accuracy, currency, or completeness of external government resources. Policies change frequently. Always verify the latest version and applicability with the relevant government agency before relying on specific documents.

Licence: Licensed under Creative Commons Attribution 4.0 (CC BY 4.0). You are free to copy, adapt and redistribute with attribution: "Source: SafeAI-Aus (safeaiaus.org)"