The influence of Artificial Intelligence (AI) is expanding in diverse domains, as seen in natural language processing tools like GPT-3, image recognition software such as Google Lens, and product recommendation engines, including Amazon’s product suggestion system. AI is gaining traction in the art world, exemplified by the sale of “Edmond de Belamy,” a portrait generated by AI, for an unprecedented $432,500 in an auction. Nonetheless, the increasing involvement of AI in creative pursuits raises copyright concerns.

When it comes to training AI models, the use of copyrighted materials is considered to be in a legal grey area. As it stands now, copyright laws do not safeguard any creation that is wholly generated by AI, regardless of whether it stemmed from a human-crafted text prompt. While fair use laws permit the use of copyrighted material under certain conditions without the owner’s permission, the ongoing legal disputes could disrupt this status quo and bring uncertainty in the future of AI model training.

Undoubtedly, the advent of generative AI has revolutionized our lifestyle, labor practices, and artistry output within a mere few months. In turn, the inundation of AI-fabricated written works, pictures, and tunes, alongside the mechanisms through which they were created, has stimulated a plethora of intricate legal inquiries. These challenge our understanding of ownership, fairness, and the core foundation of innovation.

Can AI-generated art be copyrighted?

The issue of whether AI-generated art can be protected under copyright laws has been a contentious topic, with various opinions and viewpoints. The U.S. Copyright Office has taken the position that creations made by non-human entities, including machines, are not eligible for copyright protection. Consequently, the product of a generative AI model cannot be considered copyrightable.

The fundamental challenge lies in the way generative AI systems operate. These models learn by identifying and replicating patterns found in data. Thus, the AI system must first learn from human creations to produce output such as written text or images. For example, if an AI-generated image resembles the art of Japanese artist Yokoyama Taikan, it would have been trained using actual pieces of art created by the human artist. Similarly, to generate written content in the style of J. K. Rowling, the AI system would need to be trained with words written by J. K. Rowling.

However, according to current U.S. copyright law, these AI systems – which encompass image and music generators, as well as chatbots like ChatGPT – cannot be regarded as the creators of the content they produce. Instead, their outputs result from a culmination of human-generated work, much of which is copyrighted in some form and sourced from the internet. This does not mean that AI-generated works are necessary in the public domain. Another example if a company uses AI to generate content, that company may still have proprietary rights to that content, such as a trade secret or patent.

This raises a perplexing question: how can the rapidly evolving artificial intelligence industry be harmonized with the intricate details of U.S. copyright law? This is a question that creative professionals, companies, courts, and the U.S. government are all grappling with as they navigate the complexities and nuances of AI-generated content and intellectual property laws.

Will copyright issues get tougher when humans and AI do the work together?

The issue of copyright protection for creative works resulting from collaboration between humans and machines is complex. According to the Copyright Office, if a human arranges or selects AI-generated material creatively or modifies it in a sufficiently creative way, copyright protection will only apply to the human-authored components of the work, not the AI-generated material itself. The issue of copyright protection for works created jointly by humans and machines is less clear, and registration applications must name all joint authors.

The use of generative AI for creating artistic works can also lead to copyright infringement concerns if the output shows similarities to pre-existing works on the internet. These models are often trained on existing works found online, which may lead to similarities to previous works. While there are cases where a human creatively selects or arranges AI-generated material or modifies it, resulting in copyright protection for only the human-authored aspects of the work, the situation becomes murky regarding works jointly created by humans and machines. It’s a requirement to name all joint authors, including potentially the AI, in applications for registration. It may be challenging to ascertain whether generative AI output is a derivative work or infringes upon the rights of previous authors.

Lawsuits

Getty Images has taken legal action against Stability AI, accusing the company of unlawfully copying over 12 million photos from Getty Images’ collection and utilizing them in generative AI systems without proper permission or licensing. Stability AI is not alone in facing lawsuits related to generative AI. With the launch of generative AI by numerous companies such as Microsoft, OpenAI, and GitHub, creative industries are beginning to file lawsuits over the co-opting or use of copyrighted work by AI. In addition to Getty’s case, a group of artists has also sued Stability AI, Midjourney, and DeviantArt for alleged mass copyright infringement via the use of their work in generative AI systems. These lawsuits are bringing to light the legal implications of using generative AI, which is becoming an increasingly common practice.

Legal action of collective nature was instituted against GitHub, Microsoft, and OpenAI. The motion claimed that the AI-powered coding aide GitHub Copilot infringed copyright laws by generating code derived from code licensed under open source, which is publicly accessible. Copilot provides programmers with suggestions for novel code based on their existing code in real-time. As per the legal action, Copilot’s code-generating software was trained on code that was subject to copyright, without obtaining the necessary authorization. Furthermore, the program creates new code that is akin or identical to the original work. This is the premier lawsuit to be brought involving generative AI. The case aims to attain class-action status, and if it prevails, it could potentially affect the whole AI industry and how it utilizes publicly available code for training models.

Microsoft, GitHub, and OpenAI have submitted a motion to dismiss the legal action. They argue that Copilot produces unique code and that the code generated is not merely identical copies of the data used for training.

These are some lawsuits that were filed lately involving generative AI. The resolution of the legal action and its influence on the AI industry remains unknown.

Ending Remarks

Copyright law is a fundamental aspect of protecting intellectual property and encouraging creativity. It gives creators the right to control their work’s use, distribution, and adaptation and encourages them to create more by offering them exclusive rights. Creative Commons licenses provide even more options for creators to choose the level of protection they want for their work.

As AI technology advances, it becomes increasingly involved in the creative process. With AI’s ability to generate original content and collaborate with humans, there is a growing need for a legal framework that addresses the copyright protection of collaborative works involving AI. It is crucial to strike a delicate balance between safeguarding the rights of creators and nurturing innovation and originality. It is difficult to predict the exact trajectory of copyright law as it pertains to AI-generated works. Still, it is undeniable that as AI technology becomes increasingly integrated into the creative process, the legal framework governing copyright protection will undergo significant and ongoing transformation.