Guardians Of Innovation Why Human Oversight Is Essential In AI And Copyright Protection?

The collaborative synergy between AI-generated content and human imagination represents a paradigm shift along with endless opportunities in the creative process and amplifies our creative capabilities.

Artificial Intelligence (“AI”) is rapidly evolving from a humble tool to a collaborative partner in the artistic realm, fundamentally changing how we create and experience art. In terms of quality and speed of creation of artistic works – AI is far ahead when compared to average human performance. Fueled by advancements in machine learning (“ML”), immense computational power and the quality and quantity of the data economy today, the technological landscape has undergone a significant change, which is not limited to simply assisting human artists. Today, ML can generate autonomous systems that learn without explicit human programming, presenting a compelling proposition for the creative realm. The rise and eventual use of AI in every aspect of human life, ranging from the health sector to the transportation sector, has now become inevitable. The collaborative synergy between AI-generated content and human imagination represents a paradigm shift along with endless opportunities in the creative process and amplifies our creative capabilities. Copyright law aims to protect original creations and grants the creator exclusive rights to control the use and distribution of their work; however, the emergence of AI-generated work has blurred the traditional lines of authorship and ownership, challenging the very foundations of Intellectual Property Rights Laws (“IP Laws”).

AI can now do things that could have never been imagined a decade ago. AI platforms like Midjourney, Sora, etc. act like creative platforms that are used as interactive spaces where users provide creative directions and parameters. The AI then leverages this input and generates unique, novel works such as books, images, videos and even music. This creative collaboration between AI models and the human mind can create astonishing results. However, it gives rise to multiple ethical and Copyright infringement issues that need to be intricately examined, including but not limited to, issues such as use of unlicensed content such as training data, questions regarding ownership of AI-generated works, etc. Navigating these complex issues will be essential for fostering a sustainable and responsible future for Human-AI collaboration in the creative sphere.

The Indian Copyright Act, 1957 (“Copyright Act”) mandates any work to be “original”, for grant of copyright¹. The work must involve some level of creativity and not be based entirely on skill and labour. The Copyright Act in India recognises authorship of computer-generated works. This law provides that the author of any literary, dramatic, musical, or artistic work which are computer-generated is the “person who causes the work to be created.”² Since, AI tools train on data from pre-existing sources and produce output based on such underlying works available to it on the internet, such AI tools do not satisfy the requirement of ‘creativity’ required for copyright protection; thus, making human involvement or contribution essential for ‘creativity.’

The legal implications of using copyrighted material in the training data for AI models are a topic of ongoing debate. One argument suggests that human creators should be required to obtain permission from copyright holders before incorporating their work into AI training datasets to get around the problem of copyright infringement. However, this position fails to distinguish between the concept of ‘access’ and ‘use’. While an AI model might have access to vast amounts of copyrighted material on which it is trained during its training phase, this access does not automatically grant the right to utilize that material in the generation of creative outputs. It is important to note the Copyright laws in many jurisdictions continue to distinguish between fair use and infringement. The Fair Use Doctrine permits the restricted use of copyrighted material for purposes such as commentary, news reporting, criticism, teaching, scholarship, or research³. In contrast, a copyrighted work is said to be infringed when the copyrighted work is used without permission for commercial purposes or in a way that harms the market for the original work. The key distinction here is that while AI models may be trained on copyrighted data, the resulting creative outputs must not constitute an infringing work. Most countries today rely on ‘sufficient’ human contribution in the development of the AI-generated output showing the existence of a human’s own intellectual creation or contribution to the inventive concept to be eligible for copyright protection.

There is a lack of uniformity across jurisdictions regarding this issue. The United States Copyright Office recognizes copyright as “the fruits of intellectual labour… founded in the creative powers of the mind,” thus limiting copyright protection to only human beings.⁴ The US Copyright Office on February 21, 2023, held that AI generated work cannot be granted copyright protection⁵. The Court of Justice of the European Union(CJEU) has interpreted the concept of ‘originality’ on similar lines emphasizing the need for ‘personal touch,’ ‘personality’ and ‘individual expression’ in the works created, indicating an idea of author as a natural person⁶. On the other end of the spectrum, under UK Copyright law, for computer generated works, the author is taken as the person “by whom the arrangements necessary for the creation of the work are undertaken.”⁷ This inconsistency highlighted above between different jurisdictions with respect to granting copyright protections to the AI systems or humans, showcases the need for an international standard or consensus on the issue.

As can be seen, what is of paramount importance is significant human contribution. There are different ways in which humans can contribute to AI-generated works to be eligible for copyright protection and not produce infringing outputs by using AI tools. Using generic words in the prompts instead of specifically referencing well-known characters is less likely to produce infringing output, along with fine-tuning and tailoring the prompts. For example, using a prompt like “Taylor Swift as an Olympic swimmer with a gold medal around her neck” is more likely to produce infringing material, rather than a more generic prompt such as “a young female professional swimmer with a gold medal around her neck.” AI generated art can be simply used as a starting point for subsequent creative refinement by a human artist. The more the number of such creative refinements, the more chances that the resulting work will be granted copyright protection. After the image has been created using AI tools, one can verify the originality of the image by using reverse image search apps such as Google Reverse Image Search or TinEye. These tools search for similar images as the output image on the internet and help in reducing the chances of creating infringing output. Some AI tools such as Midjourney allow the user to upload a reference picture or a base picture around which the output picture is generated, in addition to the prompts inserted by the user. It is always advised not to use a pre-existing image or any copyrighted work available on the internet, unless you own the copyright in those images, as a reference image as the output image, despite having slightly different features, will be substantially similar to the reference image uploaded. The more original and unique the reference image is, the chances of the output being original is greater and the chances of copyright infringement in this scenario becomes bleak. It is also important to always read the terms of service of the AI tools carefully and whether it allows for the commercial use of output images. Most of these AI tools absolve themselves of all liabilities and hence, the liability for copyright infringement falls on the user himself. Businesses should also be cautious in uploading any type of confidential information as the terms of service clearly state that all information entered will be used by the AI models as training data for future reference.

Recently, technology has shown such exponential growth that it is difficult for humans to keep pace with it, let alone legislations regulating such AI models. Traditional copyright laws have become outdated, and it is the need of the hour to update such laws to incorporate such a technological change in the landscape. Navigating this tricky intersection of AI and artistic creation requires a nuanced approach that aims to promote innovation yet provide protection from copyright infringements. By implementing clear guidelines for human input, fostering transparent practices, and advocating for international consensus on copyright protection, we can ensure responsible use of AI tools. This collaboration holds immense potential to push the boundaries of creative exploration and democratize artistic expression. However, the same should be done within a framework of ethical and legal considerations. As the world embraces this evolving landscape, we should aim to pave the way for a future where humans and AI collaborate to create different means of artistic expression, enriching our world in ways yet to be imagined.

Disclaimer – The views expressed in this article are the personal views of the authors and are purely informative in nature.