Disney and NBCUniversal v Midjourney: Lessons for Australian businesses on AI copyright risk
Midjourney is pushing back against Disney, NBCUniversal and Warner Bros’ copyright claim by seeking discovery of the studios’ own generative AI use, arguing it may be relevant to fair use and industry practice. The discovery dispute could shape what evidence is available at trial, including internal AI policies, prompts and training data. Australian businesses should not assume US fair use principles apply here, as Australia’s narrower fair dealing regime offers far less room to manoeuvre. The case is a timely reminder that AI governance, licensing and record-keeping need attention now, not after a dispute arises.
Last year, we wrote about Disney and NBCUniversal’s landmark copyright claim against Midjourney and what it could mean for AI-generated content. The case has now taken a new turn, with Midjourney trying to shift some of the focus back onto Hollywood’s own use of AI.
The dispute is no longer confined to whether Midjourney allowed users to create images resembling famous characters. It now raises a broader and more uncomfortable question: how are the studios themselves using generative AI?
Disney, NBCUniversal and Warner Bros allege that Midjourney’s AI tools infringe copyright by allowing users to generate images of well-known characters, including Darth Vader, Shrek and other characters in which copyright subsists. Midjourney is now seeking discovery of documents relating to the studios’ own use of generative AI, including whether they have trained or used AI systems on third-party copyright materials.
How is Midjourney fighting the studios’ copyright claim?
Midjourney’s latest filing argues that if the studios are also training or using generative AI tools on third-party copyright works, that evidence may be relevant to fair use, industry practice and the studios’ claims about market harm.
In other words, Midjourney wants the studios’ own AI practices to form part of its defence. If Hollywood is already using AI for storyboarding, ideation or internal production workflows, that may weaken the argument that similar conduct by AI companies is inherently infringing.
The studios have resisted the request. Their lead lawyer has described it as a “fishing expedition” and maintains that the case is not about stopping AI technology altogether, but about stopping Midjourney from copying, displaying, performing and creating derivative works from their famous characters without authorisation
Why does this AI copyright discovery dispute matter?
The discovery fight may sound procedural, but it could have real commercial consequences. Discovery controls what evidence each side can access before trial. In AI copyright cases, that may include internal AI policies, prompts, outputs, training datasets, model development documents and records showing whether copyright works were used in AI systems.
This matters because, in the United States, a central issue remains whether training AI models on copyright works and generating outputs that resemble protected content can be defended as “fair use”. U.S. fair use considers factors including the purpose and character of the use, the nature of the copyright work, the amount taken and the effect on the market for the original work.
Even if the studios are using AI internally, that would not automatically make Midjourney’s conduct lawful. The court will still need to look closely at what was copied, how it was used, whether the outputs reproduce protected expression and whether there is harm to existing or potential licensing markets.
What does this mean for businesses using AI?
For businesses, the case is a timely reminder that AI is not just a technology or productivity issue. It also raises important questions about intellectual property, governance and risk.
A business should not assume an output is safe to use simply because a third-party AI platform produced it. If an AI-generated image, video, text or design is substantially similar to copyright material, or reproduces protected characters, artwork, branding or other creative assets, the business using that output may still face legal risk.
The dispute also shows why record-keeping matters. Prompts, outputs, internal AI policies, vendor terms, training data sources and approval processes may all become relevant if a dispute arises. What starts as an informal creative workflow can later become evidence in litigation.
Does Australian copyright law allow AI training on copyright material?
Australian businesses should be careful not to assume that U.S. fair use principles apply here. Australia does not have a general fair use defence. Instead, it has a narrower fair dealing regime, which only applies for specific purposes such as research or study, criticism or review, parody or satire, reporting news and certain other limited purposes.
In practice, this means using copyright material to train AI systems in Australia will often require permission unless a specific exception applies. Current commentary suggests developers training AI models in Australia may need licences where copyright material is reproduced for training, and that Australia does not currently have a broad text and data mining exception for AI training.
The Australian Government is considering these issues through the Copyright and Artificial Intelligence Reference Group, including possible licensing arrangements, greater certainty around AI-generated material and lower-cost enforcement options. More recently, the Federal Government has also indicated an intention to establish a new Office of AI and develop Australian AI Standards, while signalling that Australian creators should retain ownership and control over the use of their work in AI systems.
For now, Australian law remains underdeveloped in this area. There is no dedicated AI copyright regime, no broad fair use exception and limited Australian case law dealing specifically with generative AI. Overseas decisions, including the Midjourney proceedings, may therefore be persuasive, even though they will not bind Australian courts.
Practical steps for businesses
The Midjourney proceedings show how quickly an AI dispute can turn into a discovery exercise, pulling in prompts, policies and records never built with litigation in mind. Most businesses using AI carry the same exposure already. Businesses using AI should consider:
- Implement an AI use policy that addresses permitted tools, approved use cases, confidentiality, copyright and human review.
- Check vendor terms to understand what rights the business receives in AI-generated outputs and whether the vendor gives any indemnity.
- Avoid risky prompts that request protected characters, brands, artworks, musicians, film stills or “in the style of” living artists or identifiable creators without permission.
- Keep records of prompts and outputs where AI is used for commercial material.
- Review workflows across marketing, design and product to identify where AI-generated content is being used.
- Consider licences where copyright material is being used to train, fine-tune or test AI systems.
Our perspective
The Midjourney proceedings show that AI copyright disputes are becoming more sophisticated. The debate is no longer just about whether AI companies used copyright works to train their models. Courts are now being asked to consider industry practice, internal AI use, licensing markets, disclosure obligations and the commercial reality of how businesses use AI.
For Australian businesses, waiting for the law to catch up is a risky strategy. AI policies, contracts, creative workflows and IP risk controls should be reviewed now, not after a dispute arises.
Macpherson Kelley’s Intellectual Property team assists businesses with AI policies and governance, copyright risk assessments, licensing strategies and the commercialisation of technology. If your business is using, developing or procuring AI tools, we can help you understand the risks and put practical protections in place.
The information contained in this article is general in nature and cannot be relied on as legal advice nor does it create an engagement. Please contact one of our lawyers listed above for advice about your specific situation.
more
insights
When software “phones home”: The risks of software copyright infringement
When disclosure delays become liability: Lessons from Brambles, Crowley and Noumi
China’s new trade mark notification process: What international brand owners need to know
stay up to date with our news & insights
Disney and NBCUniversal v Midjourney: Lessons for Australian businesses on AI copyright risk
Midjourney is pushing back against Disney, NBCUniversal and Warner Bros’ copyright claim by seeking discovery of the studios’ own generative AI use, arguing it may be relevant to fair use and industry practice. The discovery dispute could shape what evidence is available at trial, including internal AI policies, prompts and training data. Australian businesses should not assume US fair use principles apply here, as Australia’s narrower fair dealing regime offers far less room to manoeuvre. The case is a timely reminder that AI governance, licensing and record-keeping need attention now, not after a dispute arises.
Last year, we wrote about Disney and NBCUniversal’s landmark copyright claim against Midjourney and what it could mean for AI-generated content. The case has now taken a new turn, with Midjourney trying to shift some of the focus back onto Hollywood’s own use of AI.
The dispute is no longer confined to whether Midjourney allowed users to create images resembling famous characters. It now raises a broader and more uncomfortable question: how are the studios themselves using generative AI?
Disney, NBCUniversal and Warner Bros allege that Midjourney’s AI tools infringe copyright by allowing users to generate images of well-known characters, including Darth Vader, Shrek and other characters in which copyright subsists. Midjourney is now seeking discovery of documents relating to the studios’ own use of generative AI, including whether they have trained or used AI systems on third-party copyright materials.
How is Midjourney fighting the studios’ copyright claim?
Midjourney’s latest filing argues that if the studios are also training or using generative AI tools on third-party copyright works, that evidence may be relevant to fair use, industry practice and the studios’ claims about market harm.
In other words, Midjourney wants the studios’ own AI practices to form part of its defence. If Hollywood is already using AI for storyboarding, ideation or internal production workflows, that may weaken the argument that similar conduct by AI companies is inherently infringing.
The studios have resisted the request. Their lead lawyer has described it as a “fishing expedition” and maintains that the case is not about stopping AI technology altogether, but about stopping Midjourney from copying, displaying, performing and creating derivative works from their famous characters without authorisation
Why does this AI copyright discovery dispute matter?
The discovery fight may sound procedural, but it could have real commercial consequences. Discovery controls what evidence each side can access before trial. In AI copyright cases, that may include internal AI policies, prompts, outputs, training datasets, model development documents and records showing whether copyright works were used in AI systems.
This matters because, in the United States, a central issue remains whether training AI models on copyright works and generating outputs that resemble protected content can be defended as “fair use”. U.S. fair use considers factors including the purpose and character of the use, the nature of the copyright work, the amount taken and the effect on the market for the original work.
Even if the studios are using AI internally, that would not automatically make Midjourney’s conduct lawful. The court will still need to look closely at what was copied, how it was used, whether the outputs reproduce protected expression and whether there is harm to existing or potential licensing markets.
What does this mean for businesses using AI?
For businesses, the case is a timely reminder that AI is not just a technology or productivity issue. It also raises important questions about intellectual property, governance and risk.
A business should not assume an output is safe to use simply because a third-party AI platform produced it. If an AI-generated image, video, text or design is substantially similar to copyright material, or reproduces protected characters, artwork, branding or other creative assets, the business using that output may still face legal risk.
The dispute also shows why record-keeping matters. Prompts, outputs, internal AI policies, vendor terms, training data sources and approval processes may all become relevant if a dispute arises. What starts as an informal creative workflow can later become evidence in litigation.
Does Australian copyright law allow AI training on copyright material?
Australian businesses should be careful not to assume that U.S. fair use principles apply here. Australia does not have a general fair use defence. Instead, it has a narrower fair dealing regime, which only applies for specific purposes such as research or study, criticism or review, parody or satire, reporting news and certain other limited purposes.
In practice, this means using copyright material to train AI systems in Australia will often require permission unless a specific exception applies. Current commentary suggests developers training AI models in Australia may need licences where copyright material is reproduced for training, and that Australia does not currently have a broad text and data mining exception for AI training.
The Australian Government is considering these issues through the Copyright and Artificial Intelligence Reference Group, including possible licensing arrangements, greater certainty around AI-generated material and lower-cost enforcement options. More recently, the Federal Government has also indicated an intention to establish a new Office of AI and develop Australian AI Standards, while signalling that Australian creators should retain ownership and control over the use of their work in AI systems.
For now, Australian law remains underdeveloped in this area. There is no dedicated AI copyright regime, no broad fair use exception and limited Australian case law dealing specifically with generative AI. Overseas decisions, including the Midjourney proceedings, may therefore be persuasive, even though they will not bind Australian courts.
Practical steps for businesses
The Midjourney proceedings show how quickly an AI dispute can turn into a discovery exercise, pulling in prompts, policies and records never built with litigation in mind. Most businesses using AI carry the same exposure already. Businesses using AI should consider:
- Implement an AI use policy that addresses permitted tools, approved use cases, confidentiality, copyright and human review.
- Check vendor terms to understand what rights the business receives in AI-generated outputs and whether the vendor gives any indemnity.
- Avoid risky prompts that request protected characters, brands, artworks, musicians, film stills or “in the style of” living artists or identifiable creators without permission.
- Keep records of prompts and outputs where AI is used for commercial material.
- Review workflows across marketing, design and product to identify where AI-generated content is being used.
- Consider licences where copyright material is being used to train, fine-tune or test AI systems.
Our perspective
The Midjourney proceedings show that AI copyright disputes are becoming more sophisticated. The debate is no longer just about whether AI companies used copyright works to train their models. Courts are now being asked to consider industry practice, internal AI use, licensing markets, disclosure obligations and the commercial reality of how businesses use AI.
For Australian businesses, waiting for the law to catch up is a risky strategy. AI policies, contracts, creative workflows and IP risk controls should be reviewed now, not after a dispute arises.
Macpherson Kelley’s Intellectual Property team assists businesses with AI policies and governance, copyright risk assessments, licensing strategies and the commercialisation of technology. If your business is using, developing or procuring AI tools, we can help you understand the risks and put practical protections in place.