The government wants to test artificial intelligence in Crown Courts in England and Wales, including legal assistants that could support professionals with routine research and case analysis. It is being presented as part of a wider push to cut delays, make better use of court time and free staff from administrative work.
For ordinary readers, this is not a remote legal-technology story. It is about what happens when public services start using AI in places where mistakes can have serious human consequences. A court backlog means victims, witnesses and defendants wait longer for cases to be heard. But a faster system is only useful if people can trust how it reaches decisions and organises work.
The Ministry of Justice announcement, published on 9 June 2026, says AI legal assistants will be developed and tested before any wider use in the courts. The tools are expected to help legal professionals and court staff with routine casework, including research and analysis. Judges are also planning to use a separate AI tool to identify trial-ready cases and group similar hearings together, so court resources can be used more efficiently.
That sounds sensible on paper. Courts handle large volumes of paperwork, scheduling, evidence bundles and procedural detail. AI systems can be good at finding patterns in documents, summarising material and helping people move through repetitive admin. Used carefully, they could help staff spend less time on mechanical tasks and more time on work that needs judgement.
The caution is equally important. Legal work is not just another office workflow. An AI summary that leaves out a key detail, a search result that invents a case, or a scheduling recommendation that is treated as more certain than it really is could affect real people. The Guardian reported warnings from lawyers that AI must not be used as a substitute for funding and court staff, and noted recent examples where invented legal citations had already caused trouble in separate cases.
This is the difference between AI as a helper and AI as hidden authority. A helper drafts, searches, sorts or flags issues for a qualified person to check. Hidden authority is what happens when a system’s answer is quietly accepted because everyone is busy, the interface looks confident, or nobody has time to challenge it. The second version is where public trust can start to fray.
The government’s own wording recognises some of that risk. It says the new legal-assistant technology will first be trialled in highly controlled environments with standards for safe and ethical use. That matters because a courtroom tool should be judged differently from a consumer chatbot. It should have clear limits, audit trails, privacy protections and a way for professionals to see why it has suggested something.
The practical question for the public is not whether AI appears in the justice system. It already is appearing across public services. The question is what job it is being given. If AI helps list hearings, spot missing paperwork or reduce repetitive admin, it may make the system less frustrating. If it starts shaping legal analysis without proper checking, or if it becomes an excuse to run courts with too few people, the risks grow quickly.
There is also a plain-language issue. People involved in cases should not need technical knowledge to understand when AI has been used. If a tool helps prepare a document, summarise evidence or organise a hearing, the rules around disclosure and accountability need to be clear. When something goes wrong, people should know who is responsible: the professional using the tool, the court service, the software provider, or some combination of them.
For now, the best reading is cautious optimism. The backlog is a real problem, and technology should be allowed to remove unnecessary delay where it can. But the courts are one of the places where “move fast” is the wrong instinct. Any AI trial needs published evaluation, independent scrutiny and a firm rule that human professionals remain responsible for decisions that affect rights, liberty and access to justice.
If this trial works well, most people may never notice the AI directly. They may simply see fewer delays, cleaner admin and better-organised hearings. That would be a useful result. The warning sign to watch for is different: AI being sold as a cure for problems that actually need people, funding and careful reform.