Copyright and AI Generated Content
Technology, Media & Telecoms Focus
Rasha Al Ardah Partner,Intellectual Property
Ushbah Al-AinTrainee Lawyer,Corporate Commercial
AI tools have recently gone viral, they are a hot topic and the talk across the globe, with all the opportunities and risks associated to it. AI tools have also been controversial, with many supporting the use of it and others suspecting it. Many legal issues have arisen with the use of AI tools and the increase demand in adopting it by businesses. In this article we will discuss the intellectual property issues related to AI tools and content generated by it, specifically with respect to copyright.
There are many definitions for artificial intelligence (AI), and it keeps changing with time, but in simple language it is the science of making machines do things that require human being intelligence. AI is also referred to as machine learning, which is software algorithm that relies on data fed into it, to carry on different tasks that requires intelligence. AI applications have been there for a long time now; however, recent AI tools that become available has made AI available in every aspect of our life. There are many examples of AI tools, the most popular ones are obviously ChatGPT, ChatGPT-4, Medjourney and Stable Diffusion 2 among others. Such AI tools can analyse data and generate content in the form of text, images, paintings, audio, or others based on a prompt question or instructions.
A main question when it comes to AI generated content, is whether it can be protected by copyright. Deciding on whether such content qualifies for copyright protection would determine the extent of ownership and exploitation. For example, if a company uses AI tool to generate artworks that they sell to third parties, if such artwork does not qualify for copyright protection, this means anyone can utilise the artwork created, and hence, the investment made by that company would be useless.
By applying general copyright principles, works protected by copyright, are those original works of human being. Hence, to qualify for copyright protection, a work must be original and created by a human being or there is an involvement of a human being in its creation. Taking this into consideration, we may conclude that AI generated content may not necessarily be protected by copyright, and therefore, may fall in the public domain depending on the involvement of the human being in its creation. Generally, the human involvement is assessed based on whether the human author made creative choices during the production process and these choices are reflected in the output work. This is similar to the classic test of copyright protection for photos taken by a camera for example, however, with AI generated content, it could be more difficult to make this assessment.
Copyright ownership of AI content has already made its way to Courts, and it has been questioned. For example, in the US, the Copyright Office has concluded that images in a graphic novel, "Zarya of the Dawn," that were created using the artificial-intelligence system Midjourney should not have been granted copyright protection, and protection would only apply to other elements of the novel that the author wrote and arranged. The matter is still under discussion, as the author of "Zarya of the Dawn," is challenging the office decision, on the grounds that the Copyright Office did not consider the creative input of the author to generate these images.
Interestingly, the Saudi Intellectual property Authority (SAIP) has recently released a draft IP Law for public consultation, which has clearly addressed the intellectual property ownership in content created by AI tools. The Draft Law provides that works generated by AI tool can only be protected if the human involvement was prominent, it further went to explain that mere AI generated content that has no human involvement falls in the public domain. This is a very positive approach as it close the door for any different interpretations on whether AI generated content would be protected; however, it leaves a wider door opened on how to assess “prominence” of human involvement in the creation of the work.
The draft law has also touched upon the ownership of such AI generated content when protection is established. The draft law states that the person who has made the necessary arrangements to reach an intellectual property resulting from artificial intelligence shall be considered the owner of the rights of such intellectual property. This law has yet to be enacted.
As AI tools rely on existing data fed into it, the question that comes to surface is whether the use of copyright protected content to train AI tools constitutes an infringement of the already existing works.
This has already been subject to question before Courts. Getty Images, which provides stock images, filed a case against Stability AI, claiming that Stable Diffusion AI image generator is making use of millions of images of Getty Image’s without authorization to train its tool. Another case was filed by a number of artists against the same AI tool, accusing Stability AI of using their works without authorization and accordingly infringing their copyright.
Copyright infringement claim is based on author’s or copyright owner’s exclusive economic rights in terms of copying as well as creating derivative works. The use of copyright protected work within AI tools is claimed to infringe such rights, as AI tools copy the existing copyright protected works, and may create a new work that could potentially be considered as a derivative work of the original work. Such acts require authorization of the copyright owner or the author.
Recently the European Union parliament discussed a draft law regulating AI, and one of the topics discussed is copyright. While the draft law did not directly address the infringement of existing works by generative AI, the draft law provides that any copyrighted material used to train artificial intelligence tool will have to be disclosed publicly. This regulation if enacted will force AI tools to disclose the sources of their data and all copyright works, which would clearly then require these AI tools to seek authorization to use such works, otherwise, they may clearly be in infringement.
Another question that arises in this area is who would be held liable for the copyright infringement. Would it be the end user, i.e. the person who made a prompt in an AI tool and used a work generated by the AI tool? Or it will always be the liability of the AI tool as an intermediary? These are valid questions, however, there are no clear answers as of yet.
While AI tools will no doubt impact our lives significantly, the use of these tools shall be made with extra caution. The lack of copyright protection and the potential copyright infringement of existing copyrighted works make the use of these tools a little risky. Legislators are trying to regulate the use of AI tools and the coming months or years will witness lots of changes in this area.
For further information, please contact Rasha Al Ardah.
Published in May 2023
