The copyright questions around AI and music have gone from theoretical to urgent. AI models trained on copyrighted music are generating new works. AI tools are creating sounds that closely resemble specific artists. And the legal framework in Australia hasn’t caught up.

I’ve spoken to two music industry lawyers, a representative from APRA AMCOS, and several artists to piece together where Australian law currently stands and what artists should be thinking about.

The Core Legal Questions

There are three main copyright questions surrounding AI and music in Australia:

Can AI-generated music be copyrighted? Under current Australian law, copyright requires a human author. A work created entirely by AI, with no substantial human creative input, likely cannot be copyrighted in Australia. This is consistent with positions taken by most jurisdictions globally.

But the reality is rarely that clean. Most AI-generated music involves human decisions — selecting parameters, curating outputs, editing results, combining AI elements with human-created elements. The question becomes: at what point does human involvement transform an AI output into a copyrightable work?

There’s no clear Australian case law on this yet. The practical advice from lawyers I spoke to: ensure that your creative process involves substantial human decision-making and document it. If you use AI tools in production, keep records of what you created versus what the AI generated.

Can AI companies train on copyrighted music? This is the big one. Companies like OpenAI, Google, and others have trained their AI models on vast datasets that include copyrighted music. Whether this constitutes copyright infringement depends on whether the training falls under fair dealing exceptions in Australian copyright law.

In Australia, fair dealing is more limited than fair use in the US. The current exceptions (research and study, criticism and review, parody and satire, reporting news) don’t obviously cover commercial AI training on copyrighted works.

APRA AMCOS and other rights organisations globally are actively pursuing this issue. The AICC (Australian Information Industry Association) has called for clarity, and the Australian Law Reform Commission has flagged AI and copyright as a priority area.

If AI generates something that sounds like an existing work, who’s liable? This is legally complex. If an AI tool generates a melody substantially similar to a copyrighted song, the question is whether the human user, the AI company, or both are liable for infringement.

Current Australian law would likely hold the human user responsible for publishing the infringing work, since they made the decision to release it. But the AI company’s liability for creating a tool that generates infringing content is an open question.

What APRA AMCOS Is Doing

APRA AMCOS has been proactive on AI issues relative to many international collecting societies. Their current position includes:

• AI-generated works that involve substantial human creative input can be registered with APRA AMCOS, subject to review.
• Works generated entirely by AI without human creative input are not eligible for registration.
• APRA AMCOS is engaging with AI companies about licensing arrangements for training data.
• The organisation is participating in international discussions about AI copyright frameworks.

The practical implication for songwriters: if you use AI tools in your songwriting process, you can still register your works with APRA AMCOS, but be prepared to demonstrate the human creative contribution.

What Artists Should Do Now

Document your process. If you use AI tools in music creation, keep records of your creative decisions. Screen recordings of your sessions, notes about what the AI generated versus what you created, and evidence of your editorial choices all help establish your role as the creative author.

Read the terms of service. AI music tools have different terms regarding ownership of generated content. Some grant you full ownership; others retain rights. Read the terms before using any tool for commercial releases.

Register your works. Continue registering all original works with APRA AMCOS and ensuring your ISRC codes and metadata are correct. Proper documentation protects you regardless of how the legal landscape evolves.

Don’t copy recognisable elements. If an AI tool generates something that sounds like an existing copyrighted work, don’t use it. The technology doesn’t understand copyright, and you’re the one who’ll face liability.

Support advocacy. APRA AMCOS, the ARIA, and artist organisations are advocating for copyright frameworks that protect creators. Support these efforts through your membership and engagement.

The International Context

Australia doesn’t operate in isolation. International developments will influence Australian law.

The EU’s AI Act includes provisions about transparency in AI training data. The US has ongoing litigation (notably the NYT v. OpenAI case and several music-specific cases) that will set precedents. The UK is balancing AI industry interests against creator rights.

Whatever frameworks emerge internationally, Australia will likely adopt compatible positions, particularly given our trade relationships and international copyright treaty obligations.

The Bigger Picture

The fundamental tension is between AI companies that want unrestricted access to training data (including copyrighted music) and creators who want to be compensated when their work is used to train commercial AI systems.

This tension won’t be resolved by existing copyright frameworks alone. New licensing models, new royalty structures, and potentially new legislation will be needed. The music industry, with its established rights management infrastructure, may actually be better positioned to navigate this than other creative industries.

The work being done by AI consultants in Sydney and other firms in this space often involves advising both technology companies and creative industry clients on these emerging legal questions. It’s a space where the technical and legal intersect in complex ways.

For now, the advice is straightforward: make music, use AI tools if they help your process, document your creative contribution, and stay informed as the legal framework evolves. The artists who understand these issues will be better positioned to protect their work regardless of how the law develops.