Authors
Read time: 7 minutes
AI has disrupted the fashion and beauty space in a number of ways. The technology is used to enable brands to develop content more seamlessly, such as developing images of models of all shapes, sizes and skin tones or generating unique patterns and designs for clothing that have not actually been physically created. Throughout the last few years, fashion houses, such as LVMH and Christian Dior, have used AI to better target trend forecasting, inventory management and consumer product recommendations.
AI in the entertainment and media sector part 2
AI is not new to the fashion industry, but more and more brands are using AI in their fashion shows, campaigns and collections. For example, French fashion brand Casablanca created AI-generated images for its Spring/Summer 2023 collection and campaign. This allowed the designers and photographers to creative direct a campaign that could only be executed through AI, including uses of UFOs, space and the desert in a “hyperrealistic painting.” During AI Fashion Week, various innovative designers displayed collections that implemented AI in their workflow, used AI-generated models or showcased the latest in AI-generated clothing. Many brands noted that AI enhances the designers’ and photographers’ artistic visions and creativity. Brands like Givenchy, Prada and Calvin Klein are now working with “fully AI models,” like Miquela Sousa (Lil Miquela) on their campaigns, which allows the brands to produce virtual photoshoots and promotional shots/videos quickly, efficiently and economically.
In other cases, brands are using AI to develop product descriptions across platforms or to facilitate interactions with digital assistants who can respond to consumers’ questions about a product and how it looks or functions. For example, the chatbot may tell a consumer that a beauty serum helps “remove fine lines” or “establish a youthful glow.” In some cases, the technology relies on consumer input, such as tools that enable virtual try-on functionality. This technology allows consumers to browse and virtually wear clothing or try on cosmetics or accessories. Consumers can do this by creating scans of their bodies in a dressing room, or via a device at home while “testing out” different eye shadows or lipsticks. Other similar tools can analyze a user’s skin type and suggest products. The technology relies on scans that enable more precise measurements to help the consumer form a realistic idea of what a product may look like on them before they actually purchase it.
The benefits of these technologies are obvious, but the challenges are more subtle. First, creators and brands will need to decide how to allocate the risk among the parties. For example, established brands may be working with new technology providers that may not be in a position to insulate them from risk via indemnification. Alternatively, technology providers in the space have to consider developing systems to enable models to be trained with user data that they may not control, meaning that it may have to be deleted or de-identified post contract, or data that is subject to a consumer access, deletion or correction request. These issues certainly complicate contracting, as the parties on each side may have needs that are entirely at odds – with one needing the data and the reputation of the brand and the other trying to protect that reputation and control the use of their data.
The same technology that allows consumers to readily “try on” beauty products from the comfort of their own homes presents new issues for advertisers engaging content creators to promote their products in organic social posts. As AI technology advances, social media filters (e.g., “beauty” filters, contouring, blush, eyelash and eyeliner amplifications, etc.) are more readily available. When brands hire beauty influencers and talent to create content surrounding a beauty product, brands should have training and monitoring programs to ensure that creators are not fabricating reviews or falsely advertising the benefits of certain beauty products.
Another challenge is how these tools will fit into the privacy regulation framework, as the algorithms rely on mass amounts of data to provide the advertised functionality. On the data side, considerations include whether the data collected constitutes personal information, or more sensitive personal information, such as biometric information. Questions have arisen as to whether the scans taken by the technology identify individuals. In the United States, the collection of this type of data is regulated at the state level. Indeed, under the hallmark biometric privacy law, the Illinois Biometric Information Privacy Act, retailers offering these tools have been facing a sharp increase in litigation brought under this law related to their offering of virtual try-on tools. In 2022, many of the popular fashion brands found themselves in court, defending against claims that they used biometrics without the proper notice and consent required under the Illinois law. For example, in Shores v. Wella Operations U.S. LLC, the salon brand was accused of violating the Illinois law with its virtual hair dye try-on tool. And in Warmack v. Christian Dior, the designer brand was accused of violating the same law when it allowed users the ability to virtually try on sunglasses on their website. In that case, Dior prevailed on a defense applying a narrow statutory exemption for products in medical settings since sunglasses (including nonprescription) are FDA-regulated devices.
On the other side of the spectrum, providers may also rely on the exclusions under state laws for “publicly available information,” which is information that a consumer has made available to the public via media or information that consists of government records. See the California Civil Code, section 1798.140(v)(2). Related to privacy considerations discussed in the Data protection and privacy section, brands will also need to address data privacy issues, such as secondary use, sensitive data limitations and international data transfers.
Another challenge that fashion and beauty brands will need to consider is the accuracy of these tools. As exciting as the new technology may be, brands will need to consider what happens when users are dissatisfied with the real-life results even though they may have been pleased with what they saw during the virtual try-on. What happens if users place an order for customized clothing based on measurements obtained using an AI-powered body scan tool, but those measurements are inaccurate? These issues can be addressed in a platform’s terms of use, with disclaimers related to the accuracy of any imaging or scan, as well as return policies that allow for a return or exchange in the event the physical product does not appear as shown in the derived images.
Finally, brands must consider ownership and intellectual property rights as well. Who owns the copyright in the design of a garment generated by AI – the brand, the consumer or the AI technology provider? Alternatively, the design may not even be subject to copyright protection given the input may heavily rely on machine-coded algorithms rather than human input or creativity. These issues are discussed in more detail in the IP section, but they certainly come up in the fashion space as designs or patterns may be subject to claims of copyright.
Yet another possible challenge is to consider the target audience of these tools. Are teens likely to use these tools? And if so, brands may need to maneuver the thicket of laws such as the Age-Appropriate Code Design Act (AACD), which may be further complicated by the very nature of a black box AI system that powers these tools. The AACD, which is set to go into effect on July 1, 2024, imposes additional requirements on online services that are likely to be accessed by people under 18. This includes rigid limitations on data collection and use, additional notices and data protection impact assessments. The use of newly developed AI technology will further complicate compliance with AACD.
AI has significantly disrupted the fashion and beauty industry, revolutionizing various aspects of brand operations and customer experiences. However, challenges remain, including the allocation of risk between parties, navigating privacy regulations, ensuring the accuracy of the tools, addressing ownership and intellectual property rights and considering the impact of other regulations targeting this technology. Fashion and beauty brands must navigate these challenges to fully leverage the potential of AI while maintaining compliance and delivering exceptional user experiences.