TL;DR
Judicial review applications reached 2,630 between January and June 2026, and could hit 5,260 for the year if that rate holds, say academics Sam Guy and Joe Tomlinson in a blog published on 30 October. The judge running the Administrative Court suspects AI-assisted litigants in person are a cause. Their analysis finds the data does not yet show the expected signs, which is a useful check on a theory that is gaining recognition.
That projection would mean 36% more applications than in 2025 and 56% more than in 2024. The sharpest growth lies outside immigration and crime, in the ‘civil (other)’ group covering matters such as housing, planning and social security. On the authors’ figures, 2026 could be comfortably the busiest year in over a decade.
The judge’s theory
Minutes of June’s Administrative Court User Group record that Mr Justice Chamberlain drew attention to the increase and “suspects that the explanation behind this is that Litigants in Person are now filing claims with the assistance of artificial intelligence”. The authors, from the University of York and King’s College London, treat that as weighty because of his daily view of the court’s work. They also note the caution recorded in the minutes themselves.
Testing it
Their reasoning is simple. If AI were producing many weak claims, the courts’ filters should catch more of them. So they looked at two measures.
- Permission grant rates. These were slightly lower in 2025 and 2026 than in 2021-2024, though not meaningfully out of line with earlier years.
- ‘Totally Without Merit’ certifications. This label is reserved for hopeless claims, and its rate is consistent with the past decade.
Neither shows what the theory predicts, at least not yet. In our view, that does not disprove it. AI-assisted claims need not be weak ones, and these measures do not show which applications used a chatbot.
Part of a wider question
The debate reaches well beyond this court. We reported in May that the police complaints watchdog linked AI-drafted submissions to a rise in reviews. Rhondda Cynon Taf’s social services department has just made a similar, hedged link for its complaints, while the blog itself cites an Economist claim that AI is breaking the British state. The authors say it would be striking if a procedure as legalistic as judicial review, far from the low-cost complaints routes, were meaningfully affected.
Looking Forward
Guy and Tomlinson end on “a mixed picture” and call for continued study, warning that past caseload pressure has often ended in procedural change. In our view, that is the practical stake. Courts, tribunals and ombudsmen may redesign access rules around an assumption about AI, so the evidence for that assumption matters. UK public bodies would be wise to start recording AI use in claims now, rather than inferring it later from volumes alone.