Australian copyright opt-out proposal for AI training clashes with existing laws

Australia is weighing an opt-out system that would let AI firms train on copyrighted works unless creators explicitly object, reversing an earlier pledge to protect copyright holders.

Categorized in: AI News Creatives
Published on: Sep 18, 2026
Australian copyright opt-out proposal for AI training clashes with existing laws

The Australian government is considering an opt-out system that would let AI companies train on copyrighted creative works unless creators explicitly say no, according to a leaked proposal seen by the ABC. The shift would reverse an earlier commitment to protect copyright holders and has drawn sharp warnings from legal experts about what it means for writers, visual artists, and other professionals whose livelihoods depend on their intellectual property.

What the leaked proposal suggests

Details remain confidential, but the opt-out model appears designed to attract large AI firms - including OpenAI and Anthropic - to invest in Australian data centre infrastructure. Deputy Prime Minister Richard Marles said, "The opportunity for Australia economically is enormous in collaborating with frontier companies." Minister for Industry and Innovation Tim Ayres has denied any reduction in copyright protection, yet the existence of the discussions signals a possible retreat from the Albanese government's earlier pledge that no text and data mining exception would be introduced without compensation for rights holders.

Why opt-out clashes with Australian copyright law

Under the Copyright Act, the owner of a written work or image holds the exclusive right to decide how it is used. An opt-out system flips that burden. Creators would need to take active steps to shield their work from AI training, rather than companies being required to seek permission first. The practical problem is timing. By the time a creator discovers their material has been ingested and files an opt-out request, the data has already been absorbed into a neural network. AI models cannot "unlearn" what they have already processed.

The core issue extends beyond the taking of individual works. The more high-quality, copyright-protected material an AI system trains on, the better its outputs become. That dynamic threatens to accelerate the displacement of the very people who produced the training data in the first place.

Replacement, not just royalties

Paying creators a royalty for training use addresses only part of the problem. The larger danger is outright replacement. Generative AI has already cut into work for visual artists, graphic designers, and illustrators, pushing down wages and eliminating roles. A short film featuring an AI-generated actor debuted last year, while Spotify recently purged millions of AI-generated tracks from its platform.

The pattern repeats in other language-based fields. Translators have been shifted from primary translation work to secondary verification tasks, a downgrade that has eroded job satisfaction. In law, graduates are increasingly assigned to check AI-generated content rather than draft original documents. A recent government report identified advertising and marketing professionals among those most exposed to AI-driven employment slowdowns.

Studies cited by University of Wollongong legal scholar Dilan Thampapillai suggest readers already prefer AI-produced writing over human work in some contexts, and that law professors favoured AI-generated answers over those written by colleagues. Expanding the pool of training data without consent would likely speed up that trajectory.

Why this matters for creatives

An opt-out system would mark the first formal step toward treating Australian creative work as free raw material for AI training. The immediate risk is not a sudden collapse of the publishing or design industries. It is a steady erosion of bargaining power, as clients gain access to tools that mimic professional output at near-zero marginal cost. Creatives who want to influence the outcome have a narrow window. The government has not yet released a formal plan, and any change to copyright practice would require a consultation process. Engaging early - through industry bodies, public submissions, and direct contact with elected representatives - is the only way to ensure the burden of protecting creative work stays where the law currently places it: on the companies that want to use it.


Get Daily AI News

Your membership also unlocks:

700+ AI Courses
700+ Certifications
Personalized AI Learning Plan
6500+ AI Tools (no Ads)
Daily AI News by job industry (no Ads)