A senior judge has raised concerns about the dangers of relying on artificial intelligence for court preparations following the discovery of references to fictional legal cases in several documents.
Mr Justice Sweeting expressed these views while rejecting the appeals of two individuals contesting their extradition to Romania for crimes committed there.
During the High Court proceedings, the Crown Prosecution Service (CPS) opposed the appeals. However, it was later revealed that two legal authorities cited by the CPS were fabricated by an AI system and did not exist.
The error was brought to light before a hearing in February when it was discovered that the incorrect cases were included in a document. The CPS acknowledged the mistake, clarifying that it was unintentional and resulted from inadequate verification processes.
In a letter in March, the CPS attributed the inaccuracies to the use of artificial intelligence but admitted that the primary failure was the lack of thorough verification by the reviewing lawyer before submitting the document.
Following an internal review, the CPS found no other concerns in the 78 cases associated with the lawyer involved, indicating a low risk of recurrence. The judge confirmed that the errors did not impact the court’s decision as they were rectified before the hearing.
While accepting the CPS’s apology and assurance of no deliberate deception, Mr Justice Sweeting emphasized the importance of appropriate oversight when employing AI in legal research to prevent similar incidents in the future.
The CPS, which does not use AI for legal decisions or charging determinations, reiterated its apology and commitment to preventing such errors in the future. The judge clarified that the mistake did not affect the legal arguments or the court’s final judgment.
