This chapter examines the collision between generative AI (GenAI) model training and foundational contract law. The central claim is that GenAI companies, through their mass data-scraping bots, acquire “actual notice” of website terms of service (ToS) and are therefore legally bound, exposing them to millions of simultaneous breach-of-contract claims. However, content creators’ contract claims, which often prohibit AI training, are frequently preempted by federal copyright law. If forced to pursue a copyright infringement claim, the GenAI company’s fair use defense often prevails. This legal conflict leaves creators without effective recourse, allowing unauthorized use of their work despite clear contractual prohibitions.

Chapter 8 sets out a reform agenda for a third-generation (3G) UK Stewardship Code, grounded in the book’s normative and empirical analysis. It begins by diagnosing three core limitations of the current regime: conceptual drift; constraints on other-regarding responsibilities linked to materiality and investor duties; and persistent implementation gaps. In response, it proposes reform along two dimensions. First, it calls for a clarified, purpose-driven definition of investor stewardship – centred on a balanced, other-regarding model of enlightened stewardship. This model recognises the interdependence between financial returns and the long-term health of economic, social, and environmental systems, drawing on emerging interpretations of Section 172 of the Companies Act 2006. Second, it advocates strengthening stewardship reporting by embedding reflexivity and institutional learning. The chapter argues that investor stewardship should evolve from a compliance exercise into a credible mechanism for aligning capital with public value. Reimagined in this way, the 3G UK Code offers a forward-looking institutional response to the governance challenges of our time.

This chapter introduces the idea of tradition and considers its worth outside the law. Head-on studies of tradition, whether in the law or otherwise, are not common. The chapter argues that there are three principal attractions of tradition: (1) as what is tried and true, and therefore reliable; (2) as what is constitutive of our identity; and (3) as what reflects the excellence of our practices over long stretches of time. The remainder of the chapter introduces the plan of the book.

This chapter addresses the issue of traditionalism’s politics. It also more broadly considers the question of the politics of constitutional theories. It argues that the strict separation of politics from law that theories such as originalism purported to make are not sustainable. And, in its final parts, the epilogue suggests that what is really needed is a return to an older metaphysics of the classical legal tradition. Traditionalism can be a bridge to that older metaphysics.

This chapter considers the relationship between tradition and change. Tradition and change are often thought to be antagonists, and there certainly is a tension between them. Nevertheless, both changes from within a tradition (endogenous change) and changes from outside it (exogenous change) depend upon adaptation in the tradition or selection of a new tradition in order to be effective. This chapter selects a few legal examples to illustrate these mechanics. As to exogenous change in particular, it chooses the issue of discrimination on the basis of race and the Supreme Court’s decision in Brown v. Board of Education, which is sometimes taken to be the paradigmatic challenge to tradition in constitutional law. It then explains the relationship between reason and tradition.

Auditability is defined as the capacity of AI systems to be independently assessed for compliance with ethical, legal, and technical standards throughout their lifecycle. The chapter explores how auditability is being formalized through emerging regulatory frameworks, such as the EU AI Act, which mandate documentation, risk assessments, and governance structures. It analyzes the diverse challenges facing AI auditability, including technical opacity, inconsistent documentation practices, lack of standardized audit tools and metrics, and conflicting principles within existing responsible AI frameworks. The discussion highlights the need for clear guidelines, harmonized international regulations, and robust socio-technical methodologies to operationalize auditability at scale. The chapter concludes by emphasizing the importance of multi-stakeholder collaboration and auditor empowerment in building an effective AI audit ecosystem. It argues that auditability must be embedded in AI development practices and governance infrastructures to ensure that AI systems are not only functional but also ethically and legally aligned.

Chapter 2 traces the evolving role of the shareholder across key theories and institutional shifts in corporate governance. It begins with a historical account of shareholder governance, from entrepreneurial proprietors to passive risk-bearers, before revisiting Berle and Means’ analysis of the separation of ownership and control. It then examines how post-war managerialism gave way to contractarian theories that reframed shareholders as holders of exit rights in a market-based governance model. Legal doctrines, voting rights, and market mechanisms reinforced shareholder centrality, despite its legitimacy remained contested. The chapter turns to the rise of institutional investors in the UK since the 1970s, marking a shift in the locus and exercise of shareholder power. Through this lens, it interrogates the normative assumptions underpinning shareholder governance and revisits the meaning of ownership and control in an age of financial intermediation. It sets the stage for reimagining investor stewardship not as a mere extension of agency theory, but as a form of institutionalised accountability, embedded in systems of power, responsibility, and public purpose.

This chapter challenges the prevailing narrative that the release of open-source models constitutes the “democratization” of generative AI (GenAI). Arguing that genuine democratization requires far more than mere access, the analysis posits that openness is a necessary but insufficient condition. Current initiatives often create a “techno-utopian mirage,” benefiting only those who already possess the requisite expertise, computational power, and economic resources. The concentration of GenAI use in wealthy, English-speaking nations reveals a digital and linguistic divide, while the industry’s reliance on a workforce of low-wage global laborers highlights exploitative practices antithetical to democratic ideals. Furthermore, companies often protect proprietary trade secrets like reinforcement learning from human feedback (RLHF) methods, which are central to model quality. Until these systemic issues are addressed, “democratization” will remain a marketing slogan rather than a meaningful reality.

This chapter introduces the book’s central theoretical framework: the dual state. Building on and extending Ernst Fraenkel’s classic concept, it offers a new explanation for the persistence of judicial independence in regimes hostile to liberal norms. The chapter argues that a stable dual state depends on containing prerogative interventions through centralized discipline and restricting normative jurisdiction through dejudicialization – both of which require a high degree of political consolidation. By reframing the autonomy-control relationship as a function of regime strength, this reconceptualization resolves a long-standing puzzle in the study of judicial politics under authoritarianism. While grounded in the Chinese case, the framework offers broader insights into the institutional logic of authoritarian legality across regimes.