Where complexity meets clarity - Visual Legal

---

Generative AI is reshaping how we create — from marketing copy and logos to music, film scripts, articles, images, and even code. 

Tools like ChatGPT, DALL·E, Stable Diffusion, Midjourney, and Claude can produce content in seconds that feels fresh and original. 

But this speed and power raise serious questions about copyright, ownership, infringement, and fair compensation — questions Australian law is still grappling with.

At Visual Legal in Adelaide, we’re helping clients on both sides of this equation: creators worried about their work being scraped to train AI, businesses using AI to generate content, and developers building or customising AI tools. 

Here’s the current legal landscape.

Australian Copyright Law – Human Authorship Is Still the Core Rule

Under the Copyright Act 1968 (Cth), copyright protects original literary, dramatic, musical, and artistic works — but only if there is human authorship. 

The High Court has consistently held that non-human entities cannot be authors. 

In Commissioner of Patents v Thaler [2022] FCAFC 62, the Full Federal Court ruled that an AI system (DABUS) could not be named as an inventor under the Patents Act 1990 (Cth) because only natural persons qualify. 

The same principle applies to copyright: pure AI-generated content with no meaningful human input is generally not protected by copyright in Australia.

If a human meaningfully contributes — by crafting detailed prompts, selecting outputs, editing, arranging, or refining the result — copyright can subsist in the final work, and the human is the author. 

The threshold is one of degree: a rough AI draft that a person then heavily crafts or curates is likely protectable; fully autonomous AI output is not.

Training AI on Copyrighted Material – The Biggest Flashpoint

Generative AI models are trained on massive datasets scraped from the internet, often including copyrighted books, articles, images, music, film stills, and more.

Under Australian law:

• Reproducing a substantial part of a copyrighted work without permission is infringement (s 14 Copyright Act 1968 (Cth)).
• There is no broad text and data mining exemption in Australia (unlike the UK, EU, Japan, or Singapore, which allow commercial or non-commercial text/data mining under certain conditions).
• Fair dealing exceptions (research/study, criticism/review, parody/satire, reporting news — ss 40–43) are narrow and unlikely to cover large-scale commercial AI training.

In December 2023, the Commonwealth Attorney-General’s Department established a Copyright and Artificial Intelligence Reference Group to consult creatives, tech companies, and legal experts on whether new licensing schemes or exceptions should be introduced. 

As of early 2026, no broad exemption has been legislated. 

Australian creatives (including the Australian Society of Authors) argue that unlicensed scraping constitutes “the greatest act of copyright theft in history” and that creators should be compensated when their works train AI models.

Ownership of AI-Generated Content

• Pure AI output (minimal or no human input): generally not copyrightable in Australia.
• Human + AI collaboration: copyright usually vests in the human contributor(s) who exercised sufficient skill, labour, or judgment.
• Businesses using AI tools: ownership depends on contracts. If you hire a developer or use a platform, ensure agreements clearly allocate IP rights (e.g., you own outputs or receive a broad licence).

Moral Rights and Ethical Concerns

Creators also hold moral rights under Part IX of the Copyright Act — right of attribution and protection against derogatory treatment. 

AI outputs that mimic or distort an artist’s style could raise moral rights issues, even if no economic infringement occurs.

The 2023 Hollywood strikes (WGA and SAG-AFTRA) highlighted similar concerns: writers demanded AI not generate literary material or be passed off as source material, protecting residuals, credit, and compensation. 

Australian creators are raising parallel issues around attribution, fair payment, and the devaluation of human creative labour.

Defamation, Misinformation, and Other Risks

AI tools can “hallucinate” — generating inaccurate, misleading, or fabricated information. 

Businesses publishing AI-generated content risk defamation claims if false statements harm reputation, or Australian Consumer Law breaches under Schedule 2 to the Competition and Consumer Act 2010 (Cth) if misleading or deceptive conduct occurs in trade or commerce.

Balancing Innovation and Protection

Australia is watching international approaches closely:

• The EU AI Act (2024) requires transparency in training data and risk assessments for high-risk AI systems.
• The US continues to debate “fair use” in AI training cases (e.g., ongoing lawsuits against Stability AI, Midjourney, Anthropic, and others by Getty Images, The New York Times, authors, and artists).

In Australia, proposals under discussion include:

• Compulsory or collective licensing schemes to compensate creators for training data use
• Clearer guidelines on human authorship thresholds
• Transparency requirements for AI developers about training datasets

The Attorney-General’s reference group continues to consult stakeholders to find a balanced framework — encouraging innovation while ensuring creators are not unfairly disadvantaged.

What This Means for You in Practice

• Creators (artists, writers, photographers, musicians): Your work can be scraped for AI training with limited current protection. Document your creative process, consider licensing your work, and monitor emerging licensing schemes.
• Businesses using AI tools: Review contracts with AI providers or developers for ownership, infringement indemnities, and licensing rights. Avoid representing pure AI output as fully human-created.
• AI developers/startups: Be transparent about training data sources; unlicensed use of Australian copyrighted material carries infringement risk.

At Visual Legal, we help artists, creators, businesses, and tech innovators navigate AI and copyright — from enforcing rights and pursuing infringements to drafting safe AI-use agreements and advising on emerging policy. 

If you’re dealing with AI-generated content, training data concerns, ownership questions, or potential infringement, book a free consultation. 

We’ll meet in our central Adelaide office or via secure teleconference, listen to your situation, explain the current law plainly, and help you protect your interests. 

No pressure, just practical support.

Questions about AI and copyright in Australia? Drop them below or get in touch — we’re here to help.