Generative artificial intelligence technology (gen-AI) has become a tool at the hands of fashion designers, which allows for a modern and innovative way of design creation, incorporating AI-informed text and image generation tools into their creative process. The fashion world has witnessed several brands, such as Collina Strada, Revolve and Gucci, proactively utilizing gen-AI to generate new works of fashion and, particularly, designs of silhouettes and prints. The numerous advantages of using AI technology within traditional fashion design creation processes include a swifter and more accurate generation of trend-led designs that result from ‘mining’ hundreds of styles and dominant looks, complemented with sewing patterns suggestions or inspiration from other design elements. Designers’ day-to-day tasks, therefore, become optimized and fast-tracked, while they can prioritize their time in engaging with their own creative process, making adjustments and often expressing their ‘free and creative choices’ in finalizing the designs. Sections 2.i.–2.ii. investigate current fashion practices and distinguish between the pre- and post-AI era of fashion design creation, whilst enriching the discussion with elements of fashion theory. Against this backdrop, Sections 2.iii–2.iv of the article explore whether newly created fashion designs—particularly silhouettes and fabric prints—that have come into existence with the assistance of gen-AI as a tool, are capable of meeting the copyright ‘originality’ threshold as their author’s own intellectual creations, provided that they defy technical functionality. In doing so, the article draws from relevant case law from the UK and other EU countries, placing particular emphasis on the CJEU jurisprudence on originality for works of applied art. In Section 3, given that using the prior art and fashion archives as ‘inspiration’ can be taken to the next level with gen-AI’s text-and-data mining (TDM), the article further investigates the circumstances under which copyright infringement in relation to AI-generated fashion design outputs could arise. The focus is on the right of reproduction under the InfoSoc Directive 2001/29/EC1 and the UK CDPA 1988,2 as well as on transformative uses of AI-generated derivative designs. The article offers an interdisciplinary contribution on the technical underpinnings of AI models, shedding light on the extent to which commonly used gen-AI models, which have been previously trained on in-copyright content, could generate copyright-infringing outputs, whilst enhancing the understanding on the design workflow of gen-AI systems. In this assessment, the analysis highlights the importance of data volume and diversity in the training and performance of gen-AI models. This mixed approach, where technological considerations inform the legal analysis, offers a spherical overview of the circumstances under which infringement would be more or less likely, thereby informing both EU and UK policymakers on the regulation of gen-AI technology.

The article concludes that the emergence of gen-AI is a valuable complementary designing tool in the fashion sector. It is argued that designers have not given up their own intellectual creativity, but have rather enhanced their efforts and creative perspectives. At the same time, the article underlines that infringement of the right of reproduction cannot be simply attributed to the use of AI designing tools, due to the technical underpinnings and operation of the technology. In the unlikely event that infringing derivative outputs could be generated, their subsequent modification by a designer could ‘transform’ them into new expressions that reflect their ‘personal touch’ and ‘free and creative choices’, turning them into non-infringing AI-assisted designs.

Source: https://academic.oup.com/jiplp/advance-article/doi/10.1093/jiplp/jpaf045/8232563