Private Law Theory

ABSTRACT
The increasing internationalisation of artificial intelligence development has placed considerable strain on the territorial foundations upon which copyright law has long been constructed. AI systems are now routinely trained on datasets assembled across multiple jurisdictions, frequently outside the territories in which the resulting models are ultimately deployed or commercialised. This reality gives rise to a policy dilemma for the United Kingdom: whether, and if so how, domestic copyright rules ought to apply to acts of model training undertaken abroad when the outputs of those processes are made available to users within the UK. This question has been brought into sharp relief by the recent consultation issued by the UK Intellectual Property Office.

This paper contends that extending the reach of UK copyright law to regulate AI training conducted overseas would be ill-advised. It argues that such an approach sits uncomfortably with settled principles of public international law and with copyright’s well-established territorial character. Through an examination of the European Union’s AI Act, the paper demonstrates that attempts to project copyright standards beyond national borders through market-access mechanisms rest on uncertain legal foundations and may fail to achieve their intended objectives. It further argues that, given the UK in particular, an extraterritorial strategy is likely to discourage investment, constrain access to advanced AI technologies, and cut against the government’s professed commitment to a pro-innovation regulatory framework. The paper concludes that resisting extraterritorial expansion is important to preserving legal legitimacy, economic competitiveness, and the UK’s standing as a credible and internationally engaged actor in AI governance.

Shemtov, Noam, Against Extraterritorial Reach: Why the UK Should Resist Extending AI Regulation to Overseas Model Training (January 19, 2026), Queen Mary Law Research Paper No 482/2026.