By daisy yang

 

Introduction

The ever-improving advancement of generative artificial intelligence (AI) has led to an unprecedented explosion in the creation of artworks, literature, and music that are generated by machines rather than human beings. The creative industry was often thought to be irreplaceable due to the narrative that machines could not possibly have the capacity to be creative, let alone generate original work. More importantly, it has led to a range of legal grey areas, particularly in intellectual property law and areas of copyright and patent law. With the recent advancements and controversies, it is clear that generative AI is here to stay. Therefore it is vital consider how AI fits within our current legal framework and whether a new set of laws specific to AI is required.

 

What is Generative AI?

In order to truly appreciate the complex entanglement between the law and technology, it is important to understand what generative AI is. There was a common misconception that AI could not produce original work which is largely attributed to the lack of understanding between predictive AI and generative AI.

Predictive AI is focused on making predictions about what might happen in the future based on patterns it has learned from past data.[1]  It works by analysing large amounts of data and using machine learning algorithms to identify patterns and correlations that can be used to make predictions. Examples of predictive AI include recommendation engines, fraud detection systems, and weather forecasting models.[2]

Generative AI, on the other hand, is focused on creating new content or generating new ideas based on what it has learned from existing data.[3] These systems are trained used existing data such as images, audio, and text. It then uses algorithms and neural networks to analyse patterns and create new works based on those patterns. The results can be surprising and often unpredictable, which is what makes them so intriguing. It can be thought of as a more creative approach to AI that can come up with new and unique outputs based on its understanding of patterns in the data. Examples of generative AI include image and video synthesis systems, natural language generation tools, and music composition software.[4]

 

Recent Generative AI Controversies – IP and Copyright Law

Generative AI has advanced substantially in the past year, especially through text-to-image generating platforms such as Stability AI or chatbots designed to provide human-like conversations such as Chat GPT. However, so too have their equivalents in the voice space, such as ElevenLabs whose voice-cloning technology can imitate just about anyone after listening to just 30 seconds of speech.

In January earlier this year, Getty Images began a lawsuit against Stability AI, a company behind the popular text-to-image generating AI called Stable Diffusion, over alleged copyright violation. Getty Images alleged that Stability AI “infringed their intellectual property rights” and “unlawfully copied and processed millions of images protected by copyright”. This was done “to benefit Stability AI’s commercial interests and to the detriment of the content creators”. They further alleged that Stability AI “chose to ignore viable licensing options and long-standing legal protections in pursuit of their standalone commercial interests”.[5]

Similarly, in April of this year, Universal Music Group issued several copyright claims on streaming platforms such as Spotify, Apple Music, YouTube, Deezer, and Tidal over the song “Heart On My Sleeve”.[6] More specifically, the song was created by an anonymous artist, known as Ghost Writer, who claims to have used AI to clone the voices of Drake and The Weeknd to create the song which went viral on video sharing platform, TikTok. The song was streamed on Spotify 629,439 times before it was pulled. At Spotify's lowest royalty rate of $0.003 per stream, that means the song earned about USD$1,888 (NZD$3,072). [7] Universal Music Group claims that the training of generative AI using their artists’ music violates copyright law. However, Jani Ihalainen, a United Kingdom intellectual property lawyer, explained to the BBC that the law around copyright and artificial intelligence was not straightforward. He said that a 'deepfaked' voice, which does not specifically copy a performance, will “most likely not be covered [by UK copyright law] and could even be considered a protected work in its own right" as the "[c]urrent legislation is nowhere near adequate to address deepfakes and the potential issues in terms of [intellectual property] and other rights".[8]

 

How Does NZ IP and Copyright Law Hold Up?

With these rapid advancements, intellectual property (“IP”) offices across the world are faced with the issue of determining the extent to which a product produced by AI should be protected. In February 2022, the Review Board of the United States Copyright Office held on an appeal that copyright does not continue in a work created by an AI in the United States, essentially confirming human authorship as a prerequisite for copyright protection in the United States.[9] This ruling came from the second application request, made by Steven Thaler, to register an AI-generated two-dimensional artwork titled “A Recent Entrance to Paradise”. Mr Thaler’s first application was rejected by the US Copyright Office in 2019 as it “lacks the human authorship necessary to support a copyright claim”, which was later upheld upon second review in 2020.[10]

The authorship of works generated by AI has thus far not been adjudicated by a New Zealand court, nor has there been any guidance from jurisprudence on the equivalent provision in the United Kingdom’s Copyright, Designs and Patents Act 1988.[11] However, for the past 50 years, New Zealand has had a standing on the issue of protection for computer-generated works, which can be protected under the Copyright Act 1994. Under section 5(2)(a), the meaning of author is attributed to the person who creates it which can be “in the case of a literary, dramatic, musical, or artistic work that is computer-generated, the person by whom the arrangements necessary for the creation of the work are undertaken”.[12] Further, “computer-generated” is defined under section 2 as “work [that] is generated by computer in circumstances such that there is no human author of the work”.[13] Therefore, it is abundantly clear that the Act deems that ‘non-human’ works can be protected by copyright.

In saying this, it is important to note that in a 2019 review of the Copyright Right Act 1994, the Government had sought submissions on AI and copyright questions, asking “whether any problems are created by the current rules relating to computer-generated works, including in light of new technological developments such as artificial intelligence”.[14] In response to this question posed, the submissions received by the review board show that most consider the current approach to be appropriate.[15]

However, it seems that the issue of owning AI-generated work is still yet to be fully settled. So far, it has been established that an author who produces computer-generated works could be protected under the Copyright Act. Although, the creator must establish that there has been a work, it is original, and it has an author in order for copyright to apply. This is because New Zealand copyright protects the expression of a person’s ideas, such as the creation of a video character designs. If a person did not make the conscious decisions to create those characters, and they were in fact AI-generated, it could be argued that there is no author of a work in this circumstance. Therefore, with no author, there is no copyright owner. This issue highlights the importance of creators keeping an accurate record of their involvement and creative control over the output in case they must prove their authorship, and thus ownership, of the work produced.

 

Are Some Changes to New Zealand IP and Copyright Law Required?

It is evident that New Zealand is one of the few counties whose laws provide some extent of copyright protection to computer-generated work. However, there are still some legal grey areas to examine, and possibly consider creating bespoke laws for.

Arising from the recent AI controversy between Getty Images and Stability AI, the most apparent issue is whether there should be a legal framework put in place to protect copyrighted work from being used to train generative AI. A sub-issue stemming from this problem is whether works created by AI that was trained on copyrighted works as original work of authorship. Even in human-created work, we walk a thin line between being inspired by something and blatant plagiarism. So, what is the difference between a human being creating art inspired by copyrighted work and a generative AI system creating work trained on copyrighted work? One could argue that if such a system were to charge users to use it, then it would be wrong to use copyrighted work to produce such images. Yet, if the service was free, would it acceptable?

The same issues can be applied to the Universal Music Group controversy. Where would New Zealand law stand in creating music output using AI trained on someone’s real voice? While the music label may own specific songs made by the artist, do they also own their voice? As the song “Heart On My Sleeve” is alleged to be completely original, is it therefore protected by copyright law even if it was created using someone else’s voice?

 

 

Conclusion

Ultimately, it may well be that the law will have to adapt on a case-by-case basis or establish a new regulatory framework specifically tailored to the unique challenges that AI-generated works bring. This framework could involve creating a new category of intellectual property rights or establishing new licensing schemes to allow for the use of AI-generated works while still protecting the rights of the creators. As AI technology continues to push the boundaries of what is possible in the creative sphere, it is important for the law to strike a balance between encouraging innovation and creativity while protecting the rights of creators and ensuring that the benefits of AI are fairly distributed. These issues highlight the need for the law to adapt to the changing landscape of generative AI and provide guidance on how to address the unique challenges posed by AI-generated works.

[1] “What is Generative AI?” (19 January 2023) McKinsey & Company < https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai#/>.

[2] Natalia Sakovich “10 Examples of Predictive Analytics” Sams Solutions https://www.sam-solutions.com/blog/examples-of-predictive-analytics/.

[3] “What is Generative AI?” (19 January 2023) McKinsey & Company < https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai#/>.

[4] “What is Generative AI?” (19 January 2023) McKinsey & Company < https://www.mckinsey.com/featured-insights/mckinsey-explainers/what-is-generative-ai#/>.

[5] James Vincent “Getty Images is Suing the Creators of AI Art Tool Stable Diffusion for Scraping Its Content” The Verge (online ed, Washington D.C., 17 January 2023).

[6] Nichola Rutherford “Drake and The Weeknd AI Song Pulled from Spotify and Apple” BBC (online ed, London, 18 April 2023

[7] Nichola Rutherford “Drake and The Weeknd AI Song Pulled from Spotify and Apple” BBC (online ed, London, 18 April 2023

[8] Nichola Rutherford “Drake and The Weeknd AI Song Pulled from Spotify and Apple” BBC (online ed, London, 18 April 2023).

[9] Ryan Abbott Second Request for Reconsideration for Refusal to Register A Recent Entrance to Paradise (Correspondence ID 1-3ZPC6C3; SR # 1-7100387071) (Copyright Review Board, United States Copyright Office, February 2022) at 6.

[10] At 2.

[11] The argument was raised in Thaler v Comptroller-General of Patents, Designs and Trade Marks [2020] EWHC 2412 (Pat) at [49] in support of a claim for a patent to be granted in respect of an invention allegedly invested by artificial intelligence, but the Court did not address it substantively: see [49(3), (4)]. Further, note the potential for this issue to be litigated in New Zealand and elsewhere in the future given continuing, rapid advances in artificial intelligence capabilities.

[12] Copyright Act 1994, s 5(2)(a).

[13] Copyright Act 1994, s 2.

[14] Issues Paper, Review of the Copyright Act 1994, November 2018 at paragraph 151, available at https://www.mbie.govt.nz/dmsdocument/3441-review-of-copyright-act-1994-issues-paper-pdf.

[15] Review of the Copyright Act 1994: Issues Paper - Summary of submissions, August 2019 at page 25, available at Review of the Copyright Act 1994: Issues paper - Summary of submissions (mbie.govt.nz).

The views expressed in the posts and comments of this blog do not necessarily reflect those of the Equal Justice Project. They should be understood as the personal opinions of the author. No information on this blog will be understood as official. The Equal Justice Project makes no representations as to the accuracy or completeness of any information on this site or found by following any link on this site. The Equal Justice Project will not be liable for any errors or omissions in this information nor for the availability of this information.

Featured image source: Flickr