WE BUILT THIS CITY: GENERATIVE AI, COPYRIGHT, AND THE BUILT ENVIRONMENT.

• Date: 01 April 2024
• Author: Ghazarian, Talin

Introduction 1228
                I. The Rise of Generative AI and the Current Legal Landscape 1232
                 A. Generative AI 1233
                 B. Copyrighted Architectural Works 1235
                 1. The Incorporation of Architectural Works in the
                 Copyright Act 1235
                 2. How the Infringement Standard has been Applied to
                 Architectural Works 1237
                II. A Cascade of Class Actions Against Generative AI
                 Companies 1242
                 A. The Lawsuits 1243
                 1. The Initial Complaint in Andersen v. Stability AI 1243
                 2. The District Court's Initial Decision in Andersen 1245
                 B. Copyright Office Decisions 1247
                 1. The Question of Authorship as Applied to
                 AI-Generated Works 1247
                 2. How Rights are Allocated in Works Created
                 with the Assistance of Generative AI 1249
                 C. A Shift in Fair Use 1252
                 1. Andy Warhol Foundation v. Goldsmith 1253
                 2. Fair Use and Technology 1255
                III. Treading Cautiously to Protect Architects' Rights 1256
                 A. Generative AI Acutely Exposes an Unforgiving
                 Infringement Standard 1257
                 1. A Different Standard for Architecture 1257
                 2. The Importance of Preserving the Integrity of
                 Infringement Actions 1260
                 B. What to Watch for When Using Generative AI in
                 Architectural Practice 1263
                 1. AI's Limitations in Design Capabilities 1265
                 2. Can and Should AI Replace Human Creativity
                 in Fields Like Architecture? 1266
                 C. Potential Legal Protections Outside of Copyright 1269
                Conclusion 1272
                

"Architecture should speak of its time and place, but yearn for timelessness." (1)

INTRODUCTION

The Empire State Building, 30 Rockefeller Plaza, and the Freedom Tower each dot New York City's skyline and are so well-known that they are part of the city's identity. They are synonymous with past periods of ingenuity, aspiration, prosperity, and unity in the city. (2) These sites' value comes from their status as tourist attractions and commodities with a built-in cache and brand that can be bought and sold in the commercial real estate market like any other building (3) or converted to meet the city's needs. (4) How does the law protect and incentivize the design of icons so inextricably woven into the fabric and identity of a city? Owners of the buildings have well-established rights through the property law regime. (5) The intellectual property (IP) regime posits protections for architectural works through copyright and grants rights to the architects who conceive of and design buildings. (6)

Architects imagine, design, and shape cities. (7) They are the creative and practical minds that bring iconic buildings to life along with housing, office spaces, restaurants, stores, and other infrastructure that individuals interact with daily. (8) With the inclusion of buildings and architectural works in the copyright regime, in theory architects are supposed to receive a tangible benefit for their creations as an incentive to continue to create. (9) In a time when populations are swelling, (10) space is becoming more limited, (11) and the climate crisis is top of mind, (12) these incentives might be critical for architects tasked with exploring new sustainable and durable ways to design to effectively house and accommodate the world's population. (13) As architects respond to evolving crises, (14) their creations and rights have the potential to be caught in the crossfire and consequences of the rise of generative Artificial Intelligence (AI) technologies.

Where certain landmark buildings define our cities, generative AI has quickly positioned itself as the new "it" disruptor technology. (15) Generative AI is starting to define how society interacts with technology, (16) how lawmakers think through regulations in response, (17) and how otherwise-settled law in the IP space is exposed and vulnerable to technological innovations. (18) Generative AI's capacity to respond to user prompts with poems, writings, works of art, music and, other art forms has raised growing concerns within the creative community, particularly among authors and artists. (19) They worry about the construction and training of generative AI models, as well as how the outputs incorporate their proprietary works, posing a threat to their livelihoods. (20) Between 2022 and 2023, 16 lawsuits have been filed against generative AI companies. (21) The allegations largely focus on IP concerns related to infringement. (22) While as of February 2024 no architects have joined any of the class actions or raised a suit of their own, the challenges generative AI poses to the IP regime begs the question: is IP the strongest protective mechanism in the realm of architectural works? (23)

Recent Copyright Office decisions regarding authorship in generative AI-assisted works (24) and courts' anticipated decisions with respect to the IP claims in the class actions will help instruct architects and their counsel on best practices when using AI technologies. The decisions will likewise guide how and when architects might take action against an infringer that is using generative AI, how the incentives under the IP regime might be impacted, and whether there are more effective legal regimes, like contract law, to protect work product. Architectural works are part of the built environment. (25) While IP concerns can sometimes feel amorphous, (26) with the inclusion of architectural works in the Copyright Act in 1990 (27) a litigation risk exists to cities and built environments, which can stunt the pursuit of projects, growth, and development. A successful finding of infringement can result in an injunction. (28) Moreover, projects might be enjoined for the duration of the litigation process. (29) As the battle over AI plays out in the courts, in the case of architecture it would also play out in the urban and built environment, as architects and the communities they serve await decisions about their respective rights and plans.

This Note argues that the pending decisions for lawsuits against generative AI companies on the merits of the IP claims should consider broader applicability to fields, like architecture, that are beyond fine arts and literature, the focus of the current litigation. Architecture, though enumerated as a protected category in the current Copyright Act, often has thin protections in practice. (30) This leaves it vulnerable to any decision against the plaintiffs, which could further undercut the incentive structure behind copyright that strives to foster creative works while preserving a public benefit. Generative AI poses a danger to the legal protections in place for architectural work products, and an adverse decision could undermine a system that is already weak when applied to the industry.

Part I lays the foundation for understanding the technology and current legal landscape. Section I.A of this Note briefly outlines the rise of generative AI technologies. Section I.B introduces the current legal standards for copyright protections as specifically applied to architectural works. Part II then evaluates the active lawsuits and relevant decisions handed down by the Copyright Office, along with a potential defense available to generative AI companies. Section II.A outlines the claims raised against AI companies and how the cases are trending. Section II.B considers how relevant institutions have defined generative AI-created works in terms of their potential copyrightability. Section II.C evaluates recent changes and developments in fair use analysis, a primary defense available in infringement suits, which have accommodated technological advancements. Part III focuses on the relationship between architects' rights and generative AI. Section III.A exposes the deficiencies of protections available to architects under the copyright regime and how generative AI acutely threatens them. Building on Section III.A, Section III.B outlines the potential consequences for generative AI in architectural practice. Section III.C queries whether viable alternatives to IP protection for architects exist through other legal means available to curb the impacts of generative AI and related litigation to avoid risks to architects' integrity and interest in the work, and risks to the preservation and further development of cities.

I. THE RISE OF GENERATIVE AI AND THE CURRENT LEGAL LANDSCAPE

As generative AI use is on the ascent, so too are the complications this new technology and its applications present to existent legal frameworks, most notably IP doctrine. (31) This Part gives a basic introduction to AI technology's capabilities. It then focuses on the rationales, principles, and protections of copyright doctrine as applied to architectural works. It is crucial to understand the technology and the settled legal practices to see how the class actions might be resolved and what impact they could have on an industry like architecture that straddles the line of both creative and practical.

A. Generative AI

Generative AI is a category within AI technologies that encompasses tools for content creation, which includes text, images, music, audio, and videos. (32) Some form of generative AI has existed since the 1960s with the advent of chatbots. (33) For example, ChatGPT, developed by OpenAI, takes a user prompt and generates text in response, drawing connections from the large repository of texts, books, articles, and code used to train it. (34) The generative AI models use neural networks. (35) When a user enters a prompt, the neural networks identify patterns within the mass of data available in the training bank to form connections and generate new content in seconds. (36) Many industries both welcome and caution the advent of such a technology. The fashion industry has recognized generative AI's capacity to solve for waste problems, tailoring, and even manufacturing concerns. (37) In the film industry, generative AI preserves the image and likeness of actors for use in posterity. (38) Similarly, the music industry has identified preservation capabilities in the use of generative AI technologies. (39)

Various industries have...