AI Copyright and IP Lawyers Brisbane
Whether you own the copyright in AI-assisted work depends on how much of you is in it. EAGLEGATE advises creators, developers and businesses on AI copyright ownership, IP protection and enforcement across Brisbane and Queensland.
The copyright question for any AI project comes back to the same point: under the Copyright Act 1968 (Cth), copyright requires a human author who has contributed independent intellectual effort. Artificial intelligence has no legal status and cannot be an author. Work produced solely by AI — with minimal human creative involvement — is unlikely to attract copyright protection in Australia. Work produced by a human using AI as a tool — where the human has exercised genuine creative judgement — is more likely to be protected. The distance between these two situations, and where a specific project sits on that spectrum, is not always clear, and Australian law has not yet settled it.
This is not a hypothetical issue. Businesses that have built AI-generated marketing content, software code, design assets, data models, or product documentation using AI tools may be operating under an assumption of ownership that Australian copyright law does not yet clearly support. EAGLEGATE advises on the copyright position of AI-assisted work, the IP protection strategy that is available in the current legal environment, and the commercial steps that maximise the defensibility of ownership claims as the law evolves.
Under Australian law, the more of you there is in what your AI built, the stronger your IP position. Establishing that position — and protecting it — starts with understanding exactly where you sit on the spectrum.
Our Expertise
The Human Authorship Spectrum
Australian copyright law requires that a work must be the product of independent intellectual effort from a human author. Activities that may constitute sufficient human contribution include: research, analysis, selection, arrangement, refinement, and creative direction applied to AI outputs. A person who provides a detailed and specific prompt that directs the AI’s output, selects among multiple AI-generated options, substantially edits and reworks the output, and combines it with their own original material is in a stronger copyright position than a person who provides a brief prompt and publishes the result unmodified. EAGLEGATE assesses the specific facts of how a project was developed and advises on the copyright position that assessment supports.
AI Platform Terms — Who Actually Owns Your Outputs
Most creators and businesses using AI tools assume they own the outputs. That assumption is grounded in the platform’s terms of service — and those terms vary significantly, are regularly updated, and frequently contain provisions that are commercially significant. Key questions EAGLEGATE examines in AI platform terms include: who owns the outputs the user generates; what licence does the platform retain; can the platform use the user’s outputs for training its models; what restrictions apply to commercial use; and what happens to ownership if the user breaches the platform’s acceptable use policies. Understanding what the platform terms actually say is the starting point — not the endpoint — of any AI IP advice.
Training Data and Copyright Risk
Building or fine-tuning a custom AI model using third-party copyright material — text, images, code, audio, or other content — without authorisation may constitute copyright infringement under section 36 of the Copyright Act 1968 (Cth). The Australian Government has explicitly rejected a proposal that would have permitted the use of copyright material to train AI systems without consent. This means businesses that have trained custom models on scraped or third-party data face a live copyright risk that must be assessed. EAGLEGATE advises on the copyright position of training data, the available licences and fair dealing provisions, and the risk management steps available.
IP Protection Strategy for AI Projects
Where AI-generated or AI-assisted outputs may not attract copyright protection, alternative IP protection strategies become important. These include: trade secret and confidential information protection for proprietary AI model weights, training datasets, and algorithms; patent protection under the Patents Act 1990 (Cth) for novel AI-implemented inventions and methods; trade mark registration under the Trade Marks Act 1995 (Cth) for the AI product brand; and contractual protection through NDAs and development agreements that restrict disclosure and assign ownership. EAGLEGATE advises on the complete IP protection strategy — not just copyright alone.
Commercialising AI IP
Commercialising an AI project — through licensing, joint ventures, investment, or sale — requires clear IP documentation. Investors and acquirers will ask: who owns the underlying IP; are there third-party IP rights embedded in the training data or model; what do the AI platform terms say; and is the IP assignable? EAGLEGATE prepares AI projects for commercialisation by establishing the IP position, identifying and remediating gaps, and producing the documentation required for due diligence.
Defending AI-Related Copyright Claims
EAGLEGATE also advises businesses and individuals defending copyright claims arising from AI projects. These include: allegations that AI training data incorporated copyright material without authorisation; claims that AI-generated outputs are substantially similar to a copyright owner’s work; and disputes over ownership of outputs produced under AI development contracts where the IP provisions are unclear. EAGLEGATE advises on the available defences including fair dealing provisions under the Copyright Act 1968 (Cth), contractual rights, and the substantive copyright position of the disputed works.
Our Approach
1. Assess the Copyright Position
We assess the specific facts of how the AI project was developed — the degree of human involvement, the platform terms, the training data sources, the output modification and selection process — and advise on the copyright position that assessment supports.
2. Identify and Close IP Gaps
We identify gaps in the IP framework — undocumented ownership, unclear platform term implications, unaddressed training data risks — and advise on the steps to close them.
3. Build the Protection Strategy
We build the IP protection strategy appropriate to the project — copyright, trade secrets, patents, trade marks, contractual protection — drawing on EAGLEGATE’s full IP capability.
4. Enforce or Defend
Where IP rights are infringed, EAGLEGATE pursues enforcement. Where claims are brought against the client’s AI project, EAGLEGATE advises on the defence. Both positions are handled as a matter of standard practice.
Why Choose EAGLEGATE
IP Depth — Doyle's Guide Recognised
EAGLEGATE has been recognised by Doyle’s Guide for Intellectual Property in Queensland continuously since 2020. Copyright ownership and enforcement — the central questions in AI IP matters — are core EAGLEGATE practice areas, not extensions of a general commercial law practice.
Technical Understanding
Our founder’s engineering background means EAGLEGATE understands the relationship between AI systems and their outputs at a technical level — what the model does, what the human does, and where the creative contribution actually sits. That understanding directly informs the copyright analysis.
Current on Evolving Law
The Australian Government’s Copyright and AI Consultation Paper is active. The Copyright Act 1968 (Cth) is under reform consideration. EAGLEGATE advises on the current law and its trajectory — not on positions that predate the current reform process.
Brisbane & Queensland
We advise AI creators, developers, investors and businesses on AI copyright and IP across Brisbane, Queensland, and nationally.
Our Insights
- Does Australian copyright law protect AI-generated work?
Only where there is sufficient human authorship. The Copyright Act 1968 (Cth) has not been amended to address AI specifically, and Australia has no AI-specific copyright legislation. The existing requirement of human authorship — independent intellectual effort applied by a human — remains the threshold. Purely machine-generated outputs without meaningful human creative contribution are unlikely to be protected. The Australian Government is currently consulting on reform of this position.
- Who owns AI-assisted work where I directed the AI and edited the outputs?
This depends on the degree and nature of the human contribution. Where a person has exercised genuine creative judgement — providing specific and detailed direction, selecting among alternatives, substantially editing and reworking outputs, and combining them with original material — the work is more likely to attract copyright protection as the product of that person’s intellectual effort. The Copyright Act 1968 (Cth) does not define the minimum threshold precisely, and each situation must be assessed on its specific facts.
- Can I train an AI model on third-party copyright material?
Not without authorisation. Australia has no text and data mining exception to copyright. A proposal to introduce one — floated by the Productivity Commission in its August 2025 interim report Harnessing data and digital technology — was rejected by the Australian Government in October 2025. Using copyright material to train an AI model without the copyright owner’s licence is likely to constitute copyright infringement under section 36 of the Copyright Act 1968 (Cth). Businesses that have trained custom models on scraped or third-party data should obtain legal advice on their position.
- Who owns my AI outputs — me or the platform I used?
Typically you do — but the platform’s terms of service govern this, not a general assumption. Platform terms vary significantly and are regularly updated. Some reserve rights to use your outputs for model training. Some restrict commercial use in specific ways. Some contain provisions that affect ownership if their acceptable use policies are breached. EAGLEGATE reviews platform terms as part of any AI IP advice.
- What IP protection is available if copyright does not apply to my AI outputs?
Several alternatives may be available: trade secret and confidential information protection for proprietary AI processes, weights, and datasets; patent protection for novel AI-implemented inventions or methods under the Patents Act 1990 (Cth); trade mark registration for the AI product brand; and contractual protection through NDAs and restricted access arrangements. EAGLEGATE advises on the combination most appropriate to the specific project.
- What if someone copies what my AI project produced without permission?
If the output is copyright-protected, infringement claims are available under the Copyright Act 1968 (Cth) including injunctions, damages, and account of profits. If the output is not clearly copyright-protected, other remedies may be available through trade secret law, passing off, or misleading and deceptive conduct under the Competition and Consumer Act 2010 (Cth). EAGLEGATE advises on the available enforcement options based on the specific IP position of the project.
- I have received a copyright claim relating to my AI project — what should I do?
Seek legal advice immediately — before making any admissions, ceasing any activities, or responding to the claimant’s lawyers. EAGLEGATE advises on the strength of the copyright claim, the available defences, and the most effective strategy for responding. Claims relating to AI training data, AI-generated outputs, and AI platform use are all areas EAGLEGATE advises on from the defendant position.
- Can I obtain a patent for an AI-based invention in Australia?
Potentially, yes. Under the Patents Act 1990 (Cth), a patent may be available for a novel and inventive AI-implemented method, process, or system. AI-related patent applications face specific examination challenges at IP Australia, particularly around the requirement for a human inventor and the patentable subject matter criteria for computer-implemented inventions. The key is demonstrating that the AI is implementing a specific, novel technical solution to a specific technical problem — not simply automating a method that is obvious to apply to software. EAGLEGATE advises on patent strategy for AI inventions in conjunction with registered patent attorneys.
- Is AI-generated content that makes false statements about a person defamatory?
This is an emerging and genuinely unsettled area of Australian law. Defamation under the Defamation Act 2005 (Qld) requires publication of matter that is defamatory of the plaintiff. Where an AI system generates and publishes false statements about an identifiable person — a phenomenon sometimes called “AI hallucination” — the publisher of that content may be liable in defamation, even if the false statement was generated by AI rather than authored by a human. The operator who deployed the AI and the platform through which it was published are the most likely defendants. Managing this risk through output review processes, appropriate disclaimers, and robust content governance is a practical necessity for any AI project that generates statements about real people.
- What happens to copyright in AI-assisted work when a key contributor leaves the project?
When a co-founder, employee, or contractor who contributed to an AI-assisted creative project leaves, the copyright position of their contributions requires careful assessment. Under section 35 of the Copyright Act 1968 (Cth), copyright in work created by an employee in the course of their employment vests in the employer. Copyright in work created by an independent contractor vests in the contractor absent an express written assignment. Where contributions to AI-assisted work are not clearly documented and ownership not properly assigned, the departure of a key contributor can create an ownership dispute that is both expensive to resolve and damaging to commercialisation plans. EAGLEGATE advises on securing assignments before contributors leave and on resolving ownership disputes that arise after they do.
- What should I do if my AI is generating outputs that look substantially similar to a competitor's copyright work?
Stop using those outputs commercially until the IP position is assessed. If an AI model has been trained on or exposed to a competitor’s copyright material, its outputs may reproduce substantial portions of that material in a way that constitutes copyright infringement under section 36 of the Copyright Act 1968 (Cth). This is a real and documented risk with generative AI systems that have been trained on publicly available content that includes copyright-protected works. EAGLEGATE assesses the specific outputs, the training data provenance, and the available defences, and advises on the steps required to remediate the position.
General information only. Not legal advice. For advice specific to your situation, contact EAGLEGATE Lawyers.