The next stage of AI will be litigated in the courts

.

Artificial intelligence is no longer creeping into the wider economy. It is barrelling in.

The speed and scale of its arrival would have seemed unlikely even five years ago. Yet here we are in 2026, with large language models embedded in everyday tools for search, document drafting, coding, and decision support.

At the same time, regulatory frameworks are struggling to keep pace. As our Chair, Sharon White, set out last week in her article on AI regulation, we are seeing sharply different approaches across jurisdictions. The US has taken a hands-off approach, the EU has rushed to regulate early, and the UK is attempting to chart a middle path with principles-based regulation. Sharon describes this as an opportunity for the UK to reap the benefits of innovation while staying alive to the risks.

I agree with that assessment. But from where I sit, working at the intersection of AI and economics, what strikes me most is not just the divergence in regulatory approaches, but what happens in the gaps between them. Businesses are making adoption decisions with partial information about risk. Regulatory frameworks remain fragmented and uncertain. Business practices are evolving faster than compliance standards can keep up. When this happens, it falls to courts and tribunals to determine what constitutes reasonable conduct, sufficient safeguards, and fair assignment of liability. Litigation will not only respond to the challenges AI creates, it will play a defining role in shaping how AI is governed and trusted.