Australia considers an opt-out system for AI training on copyrighted works; lawyers argue it conflicts with the law
A leaked proposal from the Australian government envisages an opt-out system that would allow OpenAI and Anthropic to train AI on copyrighted works without consent from their creators. A legal analyst argues that this conflicts with the Copyright Act.
According to a leaked proposal obtained by broadcaster ABC, the Australian government is considering introducing an opt-out system that would allow companies such as OpenAI and Anthropic to train AI models on copyrighted works by Australians without prior consent from their creators. There is no firm plan yet, but according to statements by Deputy Prime Minister Richard Marles, the move is intended to attract major AI companies to invest in Australia, including building data centers. Minister for Industry and Innovation Tim Ayres denied that this would weaken copyright protection.
According to the legal analysis by the author of the article (a legal academic), the opt-out system is inconsistent with the Australian Copyright Act, which gives copyright owners the right to decide how their works are used — the burden of seeking consent should therefore fall on AI companies, rather than on creators who would have to actively opt out. The proposal is also described as a retreat from an earlier promise by the government led by Albanese that no text and data mining exception would be introduced for AI companies under which authors would receive compensation for the use of their works in training.
The practical problem is that once materials have been included in model training, they cannot subsequently be removed from the model, even after a later opt-out. The author points out that the key risk is not just the use of individual works, but the growing ability of AI systems to produce high-quality content through training on high-quality copyrighted works. According to the cited studies and reports, this threatens creative professions — visual artists, illustrators, translators and lawyers — where falling wages or replacement by AI are already evident. According to the article, a government report states that sectors at risk of replacement by AI (including marketing and advertising) show slower employment growth.
The author concludes that paying licensing fees to creators alone may not solve the problem, because the central concern is the risk of AI replacing human creative work, and calls for a longer and more open public consultation before any introduction of the system.
Why it matters
For Australian creators, this represents a fundamental shift in responsibility: instead of AI companies having to seek consent, authors would have to actively prevent the use of their works themselves, while material already included in a model cannot be removed. For AI companies such as OpenAI and Anthropic, the proposal signals an effort by the government to offer more favorable access to data in exchange for investments in data centers, which may influence their decisions about expanding into Australia even before the legislation is approved.
Two audiences, two different impacts
What this means
For individuals
Under the leaked proposal, Australian writers, illustrators and other creators would have to actively request that their works be excluded from AI model training themselves; otherwise, their works would be used automatically without consent or compensation.
For a business
The Australian government is considering regulatory relief (an opt-out system instead of a requirement to obtain consent) to attract investments by OpenAI and Anthropic in local data centers, which changes the legal framework for companies deciding where and how to license training data.
Risks and complianceCheck the original
Event sources
only one source so far · 1 publisher, 1 independent. We count feeds from the same owner only once.