The arrival of Artificial Intelligence (AI) has revolutionized many industries, showcasing the unparalleled power of computer applications, ranging from smart vacuum cleaners to advanced neural networks systems capable of generating text, images, and sounds that emulate human creations. As these intelligent machines continue to advance, individuals find new ways to integrate these innovative systems into their everyday activities.

The increasing use of AI systems, particularly generative AI models, is challenging the idea that only humans are capable of producing literary, musical or artistic works. As creators discover the power of AI and choose to use it to develop works of cultural and economic value, questions regarding the copyrightability of AI generated works start to emerge.

This dissertation aims to answer the question of whether AI generated works should be protected under copyright law. To achieve this, it considers the technical functioning of generative AI models such as ChatGPT, DALL-E, and Midjourney, which usually generate outputs in response to users’ prompts. This thesis also observes the historical evolution of copyright as an intellectual property right and analyses the essential requirements that a work must satisfy to obtain copyright protection. Additionally, it examines some legal cases that have been decided by courts and non-judicial bodies to address the aforementioned question. All of this is done from the perspectives of three of the most influential jurisdictions in copyright law: the United States (US), the United Kingdom (UK) and the European Union (EU).

This comparative analysis not only allows to identify similarities and differences in the approaches adopted by the US, UK and EU regarding copyright protection in the context of AI generated works, but it also makes possible to understand the reasoning behind such approaches. This research illustrates how the US and the EU reject the idea of granting copyright protection to creations produced by computer algorithms, asserting that such works do not meet their standards of originality, characterised by a strong connection between creative expression and human authorship. In contrast, the UK embraces the protection of AI generated works by having explicit provisions to regulate computer-generated works in circumstances where there is no human author involved.

This study concludes that works generated by AI systems should be protected by copyright law, either by modifying existing copyright rules or by formulating specific provisions. Such protection would encourage the use of AI systems within the creative field, thus increasing investment in the development of this type of technology. Recognising copyright in the context of computer-generated works would also ensure that the legal status of these works is clear to all stakeholders. Finally, granting copyright protection to creations generated by machine algorithms would acknowledge the relevance of original prompts provided by users to instigate outputs that would reflect such originality. Consequently, users who provide sufficiently original prompts should be entitled to copyright privileges.

Key words: Artificial intelligence, Generative AI, Neural Networks, Prompts, Copyright Law, Originality, Creativity, Human Authorship, AI Generated Works, Legal Cases, Comparative Analysis, United States, United Kingdom, European Union.