Executive summary

Artificial Intelligence (AI) is no longer a future workplace issue. It is already embedded in recruitment processes, rostering systems, performance management tools, surveillance technologies and workflow allocation across Australian workplaces.

Despite this, Australia has not enacted a standalone AI Act or any AI‑specific employment legislation. That absence should not be mistaken for regulatory neutrality. AI is already being regulated through a dense framework of existing employment, discrimination, work health and safety, privacy and consumer laws.

The Australian Government has deliberately chosen an incremental approach: strengthening existing laws, issuing voluntary governance frameworks and observing international developments rather than moving to a comprehensive AI statute. That approach places the burden squarely on employers, not regulators, to ensure that AI deployment aligns with established legal standards.

Internationally, Australia’s position diverges sharply from the European Union’s (EU) prescriptive AI Act but tracks closely with the United Kingdom (UK) and United States (US) preference for principles‑based oversight. In practice, however, the most immediate and enforceable regulation is emerging not from legislation but from Enterprise Agreements (EAs) and judicial scrutiny.

AI is already heavily regulated

Although Australia does not have AI‑specific legislation, AI in the workplace is far from unregulated. The development, deployment and use of AI systems are governed by a dense and overlapping framework of existing, technology‑neutral laws, supported by policy instruments and regulator guidance.

For employers, this means AI compliance obligations already exist across multiple legal domains.

Employment and discrimination law

AI systems used in decision‑making processes must comply with:

• the Fair Work Act 2009 (Cth) (Fair Work Act); and
• federal and state anti‑discrimination legislation.

Under the Fair Work Act, once an employee establishes that they hold a protected attribute and that adverse action occurred, the onus reverses. The employer must prove that the protected attribute played no part in the decision.

That reverse onus creates acute risk for employers relying on AI‑assisted processes where:

• outcomes are influenced by biased training datasets;
• decision‑making logic cannot be meaningfully explained; or
• human oversight is limited or absent.

Importantly, discrimination protections extend to job applicants, meaning that AI‑driven recruitment tools expose employers to liability well before an employment relationship begins.

AI does not need to be “designed to discriminate” to produce unlawful outcomes. Indirect discrimination caused by seemingly neutral criteria, especially where reproduced at scale, remains actionable.

Work health and safety

AI also engages WHS duties, and those obligations are becoming clearer and more explicit.

In February 2026, New South Wales passed the Work Health and Safety Amendment (Digital Work Systems) Bill 2025. The amendments impose a primary duty of care on employers to ensure that digital work systems – defined to include AI, algorithms and automated decision‑making tools – do not compromise worker health and safety.

The reform squarely captures:

• AI‑driven rostering tools;
• algorithmic productivity scoring;
• automated performance monitoring; and
• systems that allocate work or shifts without regard to human limits.

This materially strengthens avenues to challenge AI systems that contribute to stress, fatigue, excessive workloads or psychological harm. While the amendment is NSW‑specific, it reflects a broader regulatory trend toward treating AI as a work system rather than a neutral tool.

Other legal frameworks

While this article focuses on employment, employers should also be aware that AI is regulated through:

• privacy laws (including automated decision making transparency reforms taking effect from December 2026);
• Australian Consumer Law (including misleading conduct and unfair practices involving algorithmic decision-making);
• Online Safety Act 2021 (Cth) (industry codes and standards, which address AI-specific risks such as deepfakes and harmful content); and
• sector-specific regulations.

AI regulation in Australia exists but is fragmented.

The Government’s approach

The Australian Government has consistently stated that AI will be a critical driver of future productivity growth, and that regulation must be calibrated carefully to avoid stifling innovation, investment and adoption, particularly across businesses and workplaces.

In the absence of binding AI-specific legislation, the Government has relied heavily on policy and soft-law tools, including:

• AI Ethics Principles – voluntary principles promoting fairness, accountability, transparency, privacy, reliability and human‑centred design;
• Voluntary AI Safety Standards – providing practical governance guidance and mirroring the structure of proposed mandatory guardrails;
• NAIC Guidance for AI Adoption – streamlined governance expectations to prepare organisations for future regulation; and
• NAIC Guidance on AI-generated content – addressing transparency, labelling and watermarking.

These instruments are not legally binding, but they increasingly shape what regulators and courts will regard as reasonable governance.

The proposed mandatory guardrails

In September 2024, the Government consulted on introducing mandatory guardrails for high risk AI, covering accountability, risk management, transparency, testing and human oversight.

That proposal has not progressed and was absent from the National AI Plan released in December 2025. The decision to step back from mandatory regulation confirms a clear policy choice: voluntary standards will set expectations, while existing laws will do the enforcement work.

For employers, this means there is no “grace period”. The absence of new legislation does not reduce exposure.

What this means in practice

The regulatory posture produces a familiar pattern:

• AI is lawful to deploy, but only within existing legal limits;
• evidentiary and compliance risk sits with employers;
• WHS obligations are expanding rather than contracting; and
• EAs are filling governance gaps ahead of legislation.

Enterprise agreements

In the absence of prescriptive legislation, EAs have become one of the most effective and enforceable mechanisms regulating workplace AI.

These developments have not occurred through policy design but through bargaining dynamics, disputes over automation and increasing tribunal scrutiny.

Importantly, EA terms in relation to AI include broad definitions that are likely to capture technology already implemented and that reach well beyond new tools such as large language models.

If you would like to know more about specific industry trends or agreements, please reach out.

Key themes emerging in EAs and unions

While language differs across industries and EAs, several consistent AI-related-themes are now emerging.

Fair Work Commission and Federal Court practical guidance

The Fair Work Commission and Federal Court are signalling a consistent message: AI may assist legal work, but it does not replace professional judgment, evidentiary rigour or ethical responsibility.

Fair Work Commission

The Commission has recently published an Exposure Draft Guidance Note (Guidance Note) addressing the use of generative AI in Commission proceedings. Through this and broader case management expectations, it has reinforced long‑standing principles:

• responsibility for submissions rests with the party;
• AI does not excuse inaccuracy or mischaracterisation;
• procedural fairness must be preserved; and
• confidentiality risks are heightened in employment matters.

The draft Guidance Note introduces minimum requirements where generative AI is used in preparing documents filed in the Commission. These include:

• a requirement to disclose in the document that AI has been used;
• a requirement to verify that all content is accurate, relevant and supported by evidence and law (including checking that all case citations and authorities exist and are correctly stated);
• for legal practitioners and paid agents, inclusion of hyperlinks to all case law cited; and
• additional obligations for witness statements, including confirmation that the contents reflect the witness’s own knowledge and are true to the best of their knowledge.

The Guidance Note also cautions against providing confidential or personal information to public AI tools, reflecting broader privacy and confidentiality risks in employment matters.

AI may assist with drafting efficiency, but it does not replace professional judgment or evidentiary engagement. Failure to comply with the requirements in the Guidance Note may result in documents being disregarded, adverse cost consequences or dismissal of claims.

Federal Court

Federal courts have taken a firmer approach, issuing guidance on generative AI use following incidents involving fabricated or incorrect authority.

Key messages are consistent:

• every authority cited must exist and be verified;
• professional obligations cannot be outsourced to AI; and
• sanctions may follow where AI‑generated errors undermine the court’s process.

The judicial response to AI mirrors restraint jurisprudence: precision, proportionality and accountability remain decisive.

International context

Australia is positioning itself as a regulation adopter rather than regulation setter, closely watching overseas outcomes while relying on existing laws, voluntary standards and targeted reforms.

The EU Artificial Intelligence (AI) Act (in force 1 August 2024) represents the most ambitious regulatory model, imposing strict obligations on high‑risk AI, including employment‑related systems. Its complexity and compliance burden, particularly for SMEs, has attracted sustained criticism.

By contrast, the UK and US have avoided comprehensive legislation, instead reinforcing principles based oversight through existing regulators – an approach Australia has clearly adopted. This approach emphasises flexibility, innovation and regulator specific guidance.

Practical implications for employers

The current questions for employers are:

• What AI is already embedded in your business?
• Can its use (and controls implemented) be defended if necessary?
• What industry and peer trends are apparent (including union focus areas)?

Now is the time for employers to review AI deployment through the lens of existing legal obligations, rather than waiting for legislation that may not arrive.

If you would like to discuss the issues raised in this article, or the use of AI more broadly in your workplace, please contact the Workplace Relations & Safety team at HWLE.

This article was written by Chris Shelley, Partner, and Tanisha Chadha, Solicitor.