In a recent high-court hearing, a senior Crown Prosecution Service lawyer relied on a legal precedent that does not exist, reigniting worries that artificial-intelligence hallucinations are infiltrating the British justice system. The prosecutor, identified as Ravinder Dhillon, cited the fictitious case while arguing for the prosecution of an 11-year-old accused of raping his sister.
The citation appeared as CPS, R v J [2024] EWCA Crim 3529, a reference that court documents later proved to be imaginary. According to the prosecutor’s later explanation, the erroneous authority originated from an internet search that returned an AI-generated summary containing the fabricated details. The mistake was admitted in October, with Dhillon stating she had intended to rely on a genuine decision from fifteen years earlier.
Two senior high-court judges described the prosecutor’s justification as “unsatisfactory”, while an AI-law specialist labeled the incident a “serious failure”. The judges noted that the phantom citation did not alter the outcome of the boy’s challenge to the decision to proceed, and the case is scheduled for trial in the youth courts later this week.
The episode joins a growing list of AI-related mishaps in UK courts this year, many of which involve self-representing litigants in employment tribunals. It is only the second known CPS incident; a February admission revealed that lawyers in an extradition case had also cited nonexistent authorities, a mistake the service deemed unlikely to recur. Nonetheless, the cumulative tally continues to rise.
In June 2025 the high court urged the Bar Council and the Law Society to address AI hallucinations “as a matter of urgency”. Dame Victoria Sharp, then president of the King’s Bench Division, warned of “serious implications for the administration of justice” and indicated that misuse could trigger sanctions ranging from public reprimand to contempt proceedings or police referral.
The CPS maintains that it follows a strict AI policy that prohibits the technology from making legal determinations and requires human verification of any output. A spokesperson said the error stemmed from an internet search that produced an AI-generated summary. A tracker compiled by barrister Matthew Lee now lists 101 confirmed or suspected AI-related errors, highlighting privacy concerns and added workload for barristers checking citations.