In 2019, Professor Ginsburg delivered the Distinguished Visitor in Intellectual Property Lecture at the Faculty of Law, National University of Singapore. Titled “Fair Use in the US: Transformed, Reformed, Deformed?”, the lecture explored US caselaw applying the statutory fair use exception, highlighting its excesses and apparent rebalancing. Four and half years (and a pandemic) later, the Supreme Court has rendered decisions in two fair use cases (Google v Oracle; Andy Warhol Foundation v Goldsmith). Together, these controversies prompt inquiry whether the Supreme Court has redrawn the landscape of US fair use and copyright law, expanding fair use for commercial use of functional computer code, but narrowing it for at least some exploitations of “appropriation art.” That inquiry extends to the fair use doctrine’s potential to accommodate massive inputs of copyrighted works into databases to enable “machine learning” by artificial intelligence systems.

Under what conditions are we justified in acting? This paper considers an answer that, to my knowledge at least, goes largely unexplored in the literature. According to the answer in question, D has a justification for performing an act only if two conditions are met. First, there must be an undefeated reason for D to perform the act. Second, it must be the case that, were D to lack this reason, D would have an excuse for performing it. So understood, the conditions of justification incorporate the conditions of excuse—one satisfies the former by satisfying the latter, and by satisfying an additional condition. Justifications, simply put, are excuses plus.

Prompt-based AI creativity is redefining how expressive works are produced, posing novel challenges for copyright law. This article identifies three major obstacles standing in the way of AI copyrightability. First, the lack of meaningful human control in prompt-based AI creation undercuts the traditional requirement of human authorship. The cross-categorical nature of generative AI outputs also exposes a fundamental mismatch with copyright’s theoretical design. Second, recognising copyright in machine-created outputs risks creating unjustified windfalls for users who claim authorship without true creative contribution, undermining copyright’s incentive structure. Third, protecting AI outputs under copyright without distinction has broader creative and social consequences, including cognitive offloading, reduced authenticity, and stagnation in artistic diversity, which could erode the value of human creativity. This article recommends that future regulation preserve human-centred authorship in copyright law, implement transparency mechanisms for AI-assisted creations, and avoid overextending copyright in ways that erode the value of human creativity.

In this paper, I offer some thoughts inspired by Andrew Simester’s magisterial book, Fundamentals of Criminal Law. The starting point is Simester’s account of culpability as grounded in moral vice. The initial parts of the paper examine this account, and the language of “morality” and “vice” contained within it. Of particular concern is the legitimising role of invocations of morality and the complexities that come with that role. The latter parts of the paper raise two puzzles and examine their implications: the first is when moral judgements and legal judgements of blameworthiness come apart; the second when there is a gap between crime seriousness and individual culpability. Both puzzles allow us to tease out aspects of Simester’s theory and their implications for sentencing.

Artificial intelligence (“AI”) and other new technologies will shape, and are already shaping, the business models, management, structures and boundaries, and governance of corporations. This article examines selected developments in this area and their potential impacts on corporate law, governance, and regulation, using both theoretical and practical perspectives. The first part discusses AI and corporate leadership, focusing in particular on management structures, liability, and autonomous algorithmic entities. The second part proceeds to theorise firms in light of three specific business-related changes or phenomena induced by AI and other technologies. Finally, this part outlines selected impacts that these changes may have on corporate governance and the regulation of AI and online platforms.

Do results matter in criminal law? Many say “No” and recommend redrafting crimes in “the inchoate mode”. Others answer “Yes” and defend the division between complete and attempted offences. This paper considers a third (non-general) answer: “Sometimes”. Could it be that in some contexts results matter significantly more than in others? Murder and fraud are contrasted; while the former appears to be a “hard sell” for the inchoate mode, the latter has, at least in one major jurisdiction, passed into law and bedded down with apparent ease. Perhaps results matter in murder in a way they do not in fraud. Two contrasts are drawn, first that murder is a “horrific crime”, while fraud is not, and secondly, that the core of the wrong of kinds of fraud might be contained in the attempt. Maybe the mistake is to seek for a general answer to the question of whether results matter.

This article examines whether trade secrets law should be applied to protect “semi-public” databases with frontend data access that is open to the public. It argues that the incentive-based justification, whether grounded in the traditional rationale or reframed through an investment lens, does not provide a compelling basis for extending protection. The business efficiency rationale may arguably support trade secrets protection only where frontend access is restricted to a clearly defined and limited group of users. By contrast, when access is open to an indefinite public, ambiguous legal standards fail to mitigate inefficiencies and may even intensify the technological arms race between data holders and scrapers. Although deterrence could theoretically yield efficiency benefits, such an effect rests on a flawed assumption and risks suppressing activities that serve the public interest. Moreover, given the powerful and arguably overprotective alternatives already available to database holders, introducing trade secrets protection in this context risks further distorting the balance between private and public interests.

It is often necessary for a company to reach a compromise with its creditors, and where necessary, third parties. A useful tool for facilitating such a compromise is the scheme of arrangement, a court-controlled procedure for restructuring the relationship between the company and its members or creditors. The scheme provisions are, however, silent on whether a third-party release may be incorporated into a scheme. Since the mid 2000s, the Australian and UK courts have developed two different tests, namely the nexus test and the necessity test, to fill this statutory gap. This paper discusses policy concerns, if any, that the alternative tests have given cause to, and if the answer is ‘yes’, how these concerns should be addressed. The paper concludes that the necessity test does, and the nexus test does not, give cause for concern and that the latter should be adopted for determining scheme jurisdiction in all cases.

Simester has defended the following conclusions: (1) an omission to ϕ can cause outcome x; but (2) an omission to θ (where θ-ing would have prevented x) cannot relieve another agent of causal responsibility for x. In relation to (1), I contend that a fuller explanation of why the law should recognise omissions as causes, rather than as an independent head of responsibility-attribution for results, is required, and that any such explanation will raise questions regarding Simester’s distinctions between causal elements. Conclusion (2) follows from Simester’s view that “direct” causation is sufficient for causation of x. It will be argued that Simester has not yet made the case for preferring this model over a more familiar one, whereby an initial finding of “factual” causation is always open to being defeated by doctrines of intervening causation to ensure that responsibility for outcomes is doled out appropriately by the criminal law.

Tax jurisdiction is a legal concept, but it is fundamentally dependent on state capacity, technology and politics. The jurisdictional boundaries of the tax state are in turn crucial in delimiting its taxing power. This article considers how tax jurisdictional concepts, in particular residence, source and the location of consumption, are changing as the capability of states to tax labour, capital and consumption changes in a global digital economy. These changes are occurring through contestation in the “borderlands” of the tax state, between multiple states and non-state actors. Governments can enhance tax capability by cooperating with each other and with global intermediaries and by adopting new technologies, but also take contradictory steps to abrogate tax jurisdiction. The article illustrates the discussion with examples of tax jurisdiction for individuals as residents, workers, investors or consumers; and for corporations, including recent global developments aimed at taxation of multinational enterprises.