Jessica Bell, Kalus Kenny Intelex, Melbourne, Australia

On 15 July 2026, the Australian Government announced a new policy direction for the governance of artificial intelligence (AI). The proposed framework includes mandatory national standards for AI, regulation of data centres, and the immediate establishment of an Office of AI to design and implement these governance mechanisms. Of particular significance from an intellectual property perspective, the policy positions copyright as a central pillar of Australia’s AI governance framework, signalling a stronger commitment to protecting copyright interests.

Better copyright protection

For several years, Australia’s creative industries have expressed concern about the use of copyrighted works to train artificial intelligence (AI) models. Much of this concern centres on text and data scraping, whereby AI developers collect and process large volumes of online content for training purposes, including copyright-protected Australian works.

The Australian Government has progressively signalled a stronger policy position in response. In October 2025, it confirmed that it would not introduce a text and data mining exception to copyright infringement, thereby declining to create a statutory basis for AI developers to use Australian copyright works for training without permission.

The Government has now moved a step further. Under its proposed AI governance framework, copyright owners will retain ownership and control of their works, with future legislation intended to put that position beyond doubt. The proposed reforms would expressly prohibit AI systems from training on Australian copyright works without authorisation, reinforcing that decisions about whether a work may be used, and on what commercial terms, remain with the copyright owner.

Although no implementation date has been confirmed, the reforms are expected to be introduced as early as 2027.

Practical implications and unresolved questions

Despite the Government’s clear policy direction, several practical and legal questions remain unresolved. As the legislative framework is developed, a number of issues are likely to attract close scrutiny:

· Legacy AI models and market competition

Many existing AI models have already been trained on vast datasets that include creative works. The proposed framework therefore raises questions about competitive neutrality. If established AI developers are permitted to retain the commercial advantages of models trained before the new requirements take effect, while new market entrants must comply with more stringent obligations, the reforms may inadvertently create higher barriers to entry and reinforce the position of incumbent developers.

· Learning from works versus copying works

Copyright protects the expression of an idea, rather than the idea itself. AI systems generally learn patterns, styles and relationships from training material rather than reproducing works verbatim. As a result, questions remain as to whether the proposed reforms will adequately address outputs that closely imitate protected works through style, motifs, themes or other recognisable creative elements without amounting to copyright infringement in the traditional sense.

· International enforcement

The practical effect of the reforms may ultimately depend on their application to overseas AI developers. Many leading AI models are developed and trained outside Australia, raising questions about how the legislation will be enforced against foreign entities and whether Australian copyright owners will have effective remedies where infringing training occurs offshore.

Equally, if AI developers are required to obtain permission before using Australian copyright works for training, the reforms may accelerate the development of licensing markets. This raises further questions about how licences will be negotiated, whether collective licensing models will emerge, and whether smaller creators will have meaningful bargaining power alongside larger rights holders.

· The pace of AI development

Perhaps one of the biggest challenges is the speed at which AI technology is evolving compared with the relatively slower pace of legislative reform. By the time the new framework is implemented, AI tools and development practices may look substantially different from those that informed the policy debate. The legislation will therefore need to strike a careful balance between providing certainty for copyright owners and AI developers, while remaining flexible enough to accommodate future technological change.

So, what next?

The Government’s next steps will be critical in shaping the future of Australia’s AI landscape. The legislative and regulatory mechanisms ultimately adopted will have significant commercial implications for AI developers, particularly in relation to the training and deployment of AI models.

In introducing stronger copyright protections, the Government will need to navigate the competing objectives of protecting creators’ rights while supporting the continued growth of Australia’s AI sector. Achieving a sustainable framework will require a careful balance that provides meaningful protections for copyright owners without creating unnecessary barriers to responsible innovation and competition.

The creative industries and AI sector will undoubtedly continue to closely monitor the development of the proposed reforms and their practical impact. As the details of the framework emerge, the coming months will be critical in determining whether Australia’s approach can achieve that balance.