United Kingdom: a prosecutor invokes a court decision invented by AI

“A serious failure.” This is the observation made by Brian Flanagan, associate professor of law at Maynooth University and specialist in issues related to artificial intelligence. In October 2026, in a case revealed by The Guardian, a prosecutor from the Crown Prosecution Service (CPS) – the main public body responsible for criminal prosecutions in England and Wales – cited case law in the High Court which never existed. According to the British newspaper, this false reference was taken from a summary generated by artificial intelligence during an Internet search.

The prosecutor in question, Ravinder Dhillon, was responsible for a particularly sensitive case. The case concerned an 11-year-old boy, accused of raping his 7-year-old sister. The disagreement between the defense and the prosecution concerned whether the prosecution should continue. The child's lawyers criticized the prosecutors for not having sufficiently taken into account the observations of social workers, youth workers and a forensic psychologist.

An incident far from isolated

To justify maintaining the prosecution, the prosecutor thus relied on a judicial decision presented as a legal precedent, which turned out to be completely fictitious. In October, Ravinder Dhillon admitted his mistake. The High Court, however, rejected the appeal challenging the continuation of the proceedings, finding that this false reference had not influenced its decision. The case is now due to go to court this week.

This incident is far from isolated: according to a British register cited by The Guardian, 101 confirmed, suspected or disputed errors linked to the use of artificial intelligence in legal proceedings have been recorded in recent years. For Professor Brian Flanagan, this type of controversy “challenges the norms of trust in law, according to which we consider legal references to be valuable elements. We cannot invent them: it is unacceptable.” In response, the Crown Prosecution Service indicated that it had strict rules regarding the use of artificial intelligence. These notably prohibit entrusting these tools with legal decision-making and require human verification of the information they generate.