- Examines the recent cautionary cases of Cork and another v Smith [2026] EWHC 1199 (Ch) and Re A, B, C, D (Extension of assessment; Use of AI: hallucinations) [2026] EWFC 71 (B).
- Explores how to safeguard and/or mitigate the risks surrounding AI hallucinations.
Artificial intelligence (AI) will continue to transform the legal profession, having moved its focus from workflow digitisation to research and completing practical tasks; from data retrieval to reasoning, execution and drafting (caveat).
In a previous NLJ article (‘Case law hallucinations: Just an illusion?’ NLJ, 30 January 2026, pp17-18), I set out how modern AI systems naturally and independently perform and outthink their human-counterparts, developing ‘intelligence’ (that simulates human intelligence) thereby mastering complex, technical, and time-consuming tasks. Major AI systems developed for law rely on subscriptions for ultimate retrieval augmented generation (RAG); Thomson Reuters CoCounsel, Westlaw Precision and Westlaw Advantage, and Lexis+ with Protégé are a few key examples, and




