The Australian Federal Court Rules On AI Technology And Patents

Artificial intelligence (AI) is transforming how products are designed, software is written and problems are solved. If your team is using AI to help invent new solutions, you’re probably asking a big question: can AI be named as an inventor on a patent in Australia?

The short answer from Australia’s courts is no - at least not right now. But that doesn’t mean you can’t protect AI-assisted inventions. It just means you need to structure inventorship, ownership and your IP strategy the right way.

In this guide, we unpack the key Federal Court decisions on AI and patents, what they mean for startups and businesses, and the practical steps to protect your innovation in Australia.

What Did The Federal Court Decide About AI And Inventorship?

The issue came to a head in the well-known “DABUS” cases. A patent application listed an AI system (DABUS) as the sole inventor. Initially, a single judge of the Federal Court accepted that an AI could be an inventor. However, the Full Court of the Federal Court later overturned that decision, holding that under Australia’s Patents Act, an “inventor” must be a natural person.

In 2022, the High Court declined to grant special leave to appeal. Practically, this leaves the Full Court’s position as the law in Australia: an AI system cannot be named as an inventor.

Why did the Court take this view? In short, the Patents Act and the broader legal framework around ownership, entitlement and assignment assume a human inventor. Only a person can make an assignment, be employed, or have legal obligations and rights in the way the Act expects. Until Parliament changes the legislation, Australian patent law works on the basis that the act of inventing is attributed to a person, not a machine.

Importantly, none of this stops you from seeking patents for innovations developed with the help of AI. The key is identifying a human inventor (or inventors) who contributed to the inventive concept.

Can AI-Assisted Inventions Still Be Patented In Australia?

Yes. If people contributed to the inventive concept - for example, by defining the problem, designing prompts, selecting inputs, interpreting outputs, iterating approaches, or combining AI outputs with other insights - those individuals may be the inventors.

The test is about contribution to the inventive step, not who pressed “run.” If an AI tool was used, but a person’s intellectual contribution shaped the inventive idea, you can generally list that person as the inventor and proceed with a patent application.

In practice, focus on two things:

• Mapping the inventive contribution: Document who did what and how their input led to the inventive concept. This helps you identify the correct inventors and defend inventorship if questioned later.
• Ensuring entitlement to apply: Make sure your business is legally entitled to the invention via employment terms or a written assignment from the inventor(s).

If you’re unsure about the boundaries of inventorship where AI is involved, it’s wise to speak with an Intellectual Property Lawyer early, especially before filing.

Who Owns An AI-Assisted Invention - The Business, The Developer, Or Someone Else?

Ownership turns on inventorship and legal entitlement. Because an AI cannot be an inventor, a company can’t claim ownership directly from an AI. Instead, ownership typically flows from human inventors to the company by operation of law or contract.

Employees vs Contractors

If an employee creates an invention in the course of their employment, it’s often owned by the employer under the employment contract and company policies. If a contractor, consultant or collaborator contributes, the default position is different - they (or their entity) may own their contribution unless they’ve assigned the intellectual property (IP) to you in writing.

To avoid disputes or gaps in ownership, make sure you have the right agreements in place, including clear IP clauses in your Employment Contract and a separate IP Assignment where needed for contractors or co-developers.

Founders And Collaborators

If multiple founders contributed to the inventive concept, record those contributions and ensure the company becomes the rightful owner via assignment. Your founder documents should align with this - for example, your Shareholders Agreement should state that IP created by founders for the business is owned by the company and must be assigned if created personally.

Third-Party Tools And Training Data

Many teams rely on third-party AI platforms, models and datasets. Check licensing terms carefully. Some terms restrict commercialisation or require attribution; others may prohibit use of outputs for patent filings, or create confidentiality issues when you input proprietary data.

Before sharing anything sensitive with a partner or platform, cover your confidential information with a robust Non-Disclosure Agreement and check that the platform’s terms do not claim rights in your inputs or outputs.

Filing Strategy: How Should You Draft Patent Applications When Using AI?

AI doesn’t change the core requirements of Australian patent law, but it can affect how you present your application and evidence.

1) Identify Human Inventors Clearly

List the actual human inventors - the people who contributed to the inventive concept. Keep internal records of their contribution. If you’re using cross-functional teams (e.g. product, data science, domain experts), map involvement early so you can capture assignments from the right people.

2) Get The Entitlement Chain Right

Ensure you have written assignments from all inventors who are not your employees, and that your employment agreements capture IP created in the course of employment. Then, check that the named applicant (often your company) is properly entitled to apply, own, and enforce.

3) Satisfy Disclosure And “Best Method” Requirements

Australian patent law requires you to describe the invention clearly and disclose the best method known to you of performing the invention at the time of filing. If the “best method” involves AI methods, prompts, training parameters, or a particular model configuration, consider how much detail is needed to meet this obligation while still preserving competitive advantage.

Striking that balance is delicate: disclose enough to satisfy the legal test without oversharing trade secrets that aren’t strictly necessary for the invention. This is a common area where tailored legal advice is valuable.

4) Manage Obviousness And Supporting Evidence

As AI becomes more widespread, an examiner or opponent may argue that an AI-assisted step was obvious. Be ready to explain the non-obvious insight - for example, the problem framing, the selection of inputs, or the specific application that an ordinary skilled person would not have made at the priority date.

5) Keep Confidentiality Before Filing

Publicly disclosing your invention before filing can destroy patent rights. If you need to test, demonstrate, or discuss the invention with others, use a Non-Disclosure Agreement and carefully control access to documentation, datasets and prototypes.

6) Consider Complementary IP Protections

Patents are powerful, but they’re one part of a broader IP strategy. Protect your brand with a registered trade mark and license your software on clear terms. Your product name and logo can be safeguarded via Trade Mark Registration, while your model access and outputs can be governed by a Software Licence Agreement or EULA tailored to your business model.

Beyond Patents: Other IP And Compliance Issues With AI

Working with AI can raise broader legal questions beyond patents. Addressing them early reduces friction as you scale.

Copyright In AI Outputs

Copyright protects original works created by humans. In Australia, the law is unsettled on whether purely machine-generated works attract copyright. This uncertainty reinforces the value of patents for functional inventions, trade marks for branding, and contracts for governing access and usage rights.

If your team or contractors contribute original code, documentation or training data, ensure ownership is assigned to your company and consider a Copyright Licence Agreement where you need to grant or receive permissions.

Trade Secrets And Confidential Information

For some AI-driven innovations (e.g. secret prompts, weighting strategies, or proprietary datasets), keeping the know-how confidential can be more valuable than a patent. If you choose a trade secret approach, implement robust confidentiality controls, NDAs, access restrictions and staff training. Once publicly disclosed, trade secret protection is essentially lost.

Branding And Marketing Claims

AI-related claims (e.g., “AI-powered diagnostics”) must be truthful and not misleading under the Australian Consumer Law. If you’re offering a product to customers, your website and product collateral should include accurate, up-to-date information, and your brand should be protected with a registered trade mark.

Privacy And Data Governance

If you collect or use personal information for training or fine-tuning, you’ll need to comply with the Privacy Act. Be transparent about data practices and maintain a clear, accessible Privacy Policy. Contracts with data suppliers and customers should clearly address data usage, retention, de-identification and security.

Licensing And Platform Terms

Review the terms of any AI platform you use. Ensure you have the rights you need to commercialise outputs, file patents where appropriate, and keep your inputs confidential. When you provide access to your own AI tools or models, set the rules with a tailored EULA and platform terms to control usage, warranties, and liability.

What Legal Documents Should AI-Enabled Businesses Put In Place?

While your exact needs will depend on your product and commercial model, most AI-driven businesses benefit from having these core documents sorted:

• Non-Disclosure Agreement (NDA): Prevents confidential information (including data, model details, prompts and prototypes) from being shared or misused when speaking with partners, testers or vendors. You can start with a mutual or one-way Non-Disclosure Agreement.
• Employment Contract: Ensures IP created by employees in the course of their work is owned by the company, and sets expectations for confidentiality and acceptable use. Use a clear, modern Employment Contract with IP and confidentiality clauses.
• IP Assignment: Transfers IP from founders, contractors or collaborators to your company so the business owns the invention before filing. A written IP Assignment is essential where contributors are not employees.
• Software Licence Agreement / EULA: Governs how customers can access your AI tools, restrictions on use, payment terms, warranties and liability. A tailored Software Licence Agreement helps manage risk and set clear rules.
• Trade Mark Registration: Protects your product name and logo to build brand value and reduce confusion in the market. Consider early Trade Mark Registration for key brands.
• Privacy Policy: Explains how you collect and handle personal information, which is crucial if your product processes or stores user data. Publish and maintain an accurate Privacy Policy.
• Shareholders Agreement: Sets how founders make decisions, deal with IP and handle exits and investment. A strong Shareholders Agreement aligns incentives and reduces future disputes.
• Copyright Licence Agreement: Useful where you’re licensing content, datasets or code in or out of your business; clarifies scope, fees, rights and restrictions using a Copyright Licence Agreement.

You may not need every document on day one, but getting the core pieces in place before you demo, hire, or take on customers will make patent filing smoother and protect your commercial position.

Practical Tips For Teams Using AI In R&D

• Keep an “inventor log”: Track who contributed to the inventive concept, including the role of prompts, fine-tuning steps and human judgment. This will help with inventorship, entitlement and best method disclosure.
• Decide patent vs trade secret early: If you plan to patent, avoid public disclosure before filing. If you lean toward trade secrets, double down on confidentiality controls.
• Audit third-party terms: Check licensing for models, APIs and datasets to ensure you can file patents and commercialise outputs without restriction.
• Align contracts with your IP strategy: Employment, contractor and partner agreements should all point to company ownership of IP relevant to your products.
• Build a brand moat: Patents protect functionality, but brands drive trust. Secure names and logos with trade marks and be precise in your marketing claims.
• Plan for scale: As you move from prototype to product, revisit liability caps, indemnities, and data protection clauses in your customer and partner contracts.

Key Takeaways

• Australia’s Federal Court has confirmed that an AI system cannot be named as an inventor; inventors must be human.
• AI-assisted inventions can still be patented if a person contributed to the inventive concept and is properly named as the inventor.
• Ensure your company owns the invention via employment terms or written assignments from all inventors and contributors.
• When filing, identify human inventors, satisfy disclosure and best method requirements, and manage confidentiality before you go public.
• Complement patents with trade marks, licensing terms, NDAs and a clear Privacy Policy to create a robust, layered IP strategy.
• Get tailored advice from an Intellectual Property Lawyer if you’re unsure about inventorship, entitlement or how AI affects your filing strategy.

If you’d like a consultation on protecting AI-assisted inventions and your broader IP strategy, you can reach us at 1800 730 617 or team@sprintlaw.com.au for a free, no-obligations chat.