TL;DR
At Inner London Crown Court, HM Courts & Tribunals Service and the Ministry of Justice are testing AI software that checks criminal cases for unfinished tasks before trial, UKAuthority reported on 22 October. The pilot, starting this month, is meant to reduce delays caused by cases arriving at court unready. Staff decide what to do with every flag, and the tool plays no part in judicial decisions.
How the tool works
The AI Case Readiness Assistant reads through administrative case records and highlights gaps, such as missing details or actions that have not been completed, earlier than they might otherwise surface. Court staff then choose whether and how to act.
HMCTS has drawn tight boundaries around it. The tool does not make judicial decisions, will not decide outcomes and will not change how the law is applied. Listing cases for trial remains a matter for judges.
Oversight comes from the judiciary itself. The pilot is being run in consultation with the AI working group of the Criminal Courts Improvement Group, which Mr Justice Johnson, the Deputy Senior Presiding Judge, chairs.
What HMCTS says
Amar Morjaria, deputy director for digital delivery and AI, said the aim is to see how the software performs in a live court before any decision on wider use. “We test new tools carefully, learn from how they work in practice and only consider wider use where there is clear evidence of benefit,” he said.
He was also measured about the limits: “Technology alone will not solve the pressures facing the justice system.”
The assistant is one of several AI uses HMCTS is exploring for back-office work, alongside transcription, summarising text, anonymising data and searching within case management systems.
Looking forward
Whether the tool spreads to other Crown Court centres depends on the evidence from this trial. A useful measure will be whether flagged cases actually reach trial ready more often, not simply how many issues the software raises. The design keeps AI to administrative preparation, well away from the questions about AI making decisions raised in the Master of the Rolls’ recent remarks. That is, in our view, a sensible place for courts to start: errors in a readiness checklist are caught by staff, while errors in judicial reasoning are far harder to unwind.