The conflict between regulating AI and reaping its rewards: New Productivity Commission interim report released

The rapid rise of artificial intelligence (AI) has sparked growing debate around the unauthorised use of copyright works to train AI datasets.

Amongst the most concerned are Australia's creative and media sectors, with multiple organisations voicing their concerns. In our previous article dated 11 July 2025 , we discussed this rising controversy and analysed the current position in Australia with respect to the fair dealing defence that has been applied in the United States, and its unlikely application to the training of AI in Australia as our law currently stands.

There has also been mounting pressure on the Australian Government to introduce laws aimed at regulating AI.

The Australian Government released voluntary AI guardrails covering aspects of AI development and application in August 2024, however these guardrails have not yet become binding law.  

A proposals paper for making these guardrails mandatory was released by the Australian Government in September 2024.  There are also already mandatory requirements in place for Government agencies.

Harnessing data and digital technology: the interim report

On 5 August 2025, the Australian Productivity Commission (Commission) released an interim report titled "Harnessing data and digital technology" (Report). Submissions in response to the Report are due by 15 September 2025. The Report explores the opportunity for Australia to harness the consumer and productivity benefits of AI, while managing its risks.

In the Report, the Commission expresses concern that Australia's copyright laws are not keeping up with the rapid adoption of AI, and several policy options are suggested including amending the Copyright Act 1968 (Cth) to include a fair dealing defence that would cover text and data mining.

The Commission has also responded to the Australian Government's proposal to implement mandatory guardrails relating to high-risk AI, warning that "poorly designated regulation could stifle the adoption and development of AI and limit its benefits".¹

Australia's copyright regime and the rise of AI

One of the proposals explored by the Commission is whether a new fair dealing exception covering text and data mining is necessary.

This defence already exists in some overseas jurisdictions, including the United States (as discussed in our previous article of 11 July 2025).  It has been suggested by the Commission that this new exception could not only cover AI training, but all systems of analytical techniques that utilise machine-read material to recognise trends, patterns and other useful information.

The Commission is also seeking feedback on whether there should be no change to the current copyright laws in Australia such that copyright owners could continue to take action for copyright infringement.

Another proposal in relation to which the Commission is seeking feedback is whether licensing of copyright protected materials should be made easier, for example through facilitating licensing via collecting societies.

Rethinking the mandatory guardrails

As noted above, on 5 September 2024, the Australian Government released a proposals paper for introducing mandatory guardrails for AI in high-risk settings. The paper proposes 10 mandatory guardrails which must be observed by organisations developing or deploying high-risk AI systems. These include:²

1. Establishing, implementing and publishing an accountability process including governance, internal capability and a strategy for regulatory compliance.
2. Establishing and implementing a risk management process to identify and mitigate risks.
3. Protecting AI systems, and implementing data governance measures to manage data quality and provenance;
4. Testing AI models and systems to evaluate model performance and monitor the system once deployed.
5. Enabling human control or intervention in an AI system to achieve meaningful human oversight.
6. Informing end-users regarding AI-enabled decisions, interactions with AI and AI-generated content.
7. Establishing processes for people impacted by AI systems to challenge use or outcomes.
8. Being transparent with other organisations across the AI supply chain about data, models and systems to help them effectively address risks.
9. Keeping and maintaining records to allow third parties to assess compliance with guardrails.
10. Undertaking conformity assessments to demonstrate and certify compliance with the guardrails.

In its Report, the Commission has raised concerns about making these guardrails mandatory laws. The Commission is concerned that the guardrails would be applied to all AI technology with potential for high-risk use (which could include general purpose AI technology), irrespective of whether the risks could be better mitigated through other regulations.

The Commission also notes that a number of Australian Government agencies have begun an AI regulatory gap analysis of existing regulations including the Department of Health, Disability and Ageing, the Therapeutic Goods Administration and Treasury (which is considering whether the current Australian Consumer Law will adequately protect consumers who use AI).  

The Commission recommends that the Australian Government halt its efforts to mandate the guardrails until a comprehensive gap analysis of existing regulatory frameworks is completed.  The Commission also notes that AI specific regulation is relatively rare globally and recommends that AI specific regulations in Australia should only be "considered as a last resort when it is not feasible to adapt existing regulatory frameworks or introduce technology-neutral regulations".  

The Report also recommends that Australia continue to monitor and be involved with discussions about regulation of AI internationally with a view to adoption of internationally consistent regulations.

Privacy regulation

The Office of the Australian Information Commissioner (OAIC) considers that the Australian Privacy Act 1988 (Cth) already applies to the use of personal information to train AI models and all uses of AI that involve personal information.  Guidelines were published by the OAIC in October 2024 with respect to developing and training AI models.

The Commission also considers privacy regulation in its Report and notes that some requirements under Australia's privacy laws focus too much on prescribing certain actions or procedures rather than outcomes, which can contribute to excessive regulatory burden without improving outcomes for individuals.  

What is next for AI in Australia?

Any potential changes to the law will need to balance the need for Australia to keep up with the rest of the world and reap the benefits of AI against the need to fairly compensate copyright owners and the potential for misuse and unintended consequences leading to harm.

For assistance with respect to copyright or AI, please contact our Intellectual Property or Technology teams.

Authors

Katrina Chambers | Partner | +61 7 3338 7948 | kchambers@tglaw.com.au

Munir Abdul | Associate | +61 7 3338 7988 | mabdul@tglaw.com.au

Sophia Anderson | Law Graduate

Footnotes

¹ Australian Productivity Commission, Harnessing data and digital technology Interim report (August 2025).

² Department of Industry, Science and Resources, Safe and responsible AI in Australia (September2024).