From Pre-Action Protocol to Permission – Structuring Judicial Review grounds with AI — and avoiding merits traps
Judicial Review & AI – Part 6
Introduction: permission is the real battlefield
Most Judicial Review claims never reach a full hearing.
They fail — quietly and decisively — at the permission stage.
For litigants in person, this can feel bewildering. Everything may feel unfair. The process may have stalled. Appeals may have been ignored. And yet the court refuses permission in a few short paragraphs.
The reason is usually not lack of injustice.
It is poor framing.
This article explains:
- what the permission stage is actually testing,
- how Judicial Review grounds must be structured,
- why merits-based arguments are fatal,
- and how AI can help enforce discipline, not inflate claims.
What the permission stage is for (in reality)
Under CPR Part 54, the Administrative Court must decide whether a claim is:
- Arguable, and
- Suitable for Judicial Review.
This is not a mini-trial.
It is a filtering exercise.
Judges are asking:
- Is this a genuine public-law issue?
- Is there an alternative remedy?
- Is the claim focused and lawful?
- Is it proportionate for the High Court?
If the answer to any of these is “no”, permission is refused.
Why litigants in person struggle most at this stage
Litigants in person often:
- understand the facts deeply,
- experience the injustice personally,
- know exactly what feels wrong.
But Judicial Review does not operate on feelings.
It operates on:
- duties,
- legality,
- jurisdiction,
- restraint.
The hardest shift is moving from:
“This decision was wrong”
to
“This decision-making process was unlawful.”
AI can help enforce that shift — if used correctly.
The structure of Judicial Review grounds (what the court expects)
Judicial Review grounds are not free-form.
They are expected to follow a disciplined structure:
- The decision (or failure) challenged
- The legal duty or power
- The public-law ground
- How the duty was breached
- Why Judicial Review is appropriate
- The remedy sought
If any of these are missing or muddled, permission is at risk.
Ground 1: identifying the correct target
Your grounds must clearly identify:
- what is being challenged,
- when it occurred,
- who is responsible.
This may be:
- a refusal,
- a failure to determine,
- a procedural decision,
- or a constructive refusal.
Vague formulations (“the court has ignored me”) almost always fail.
AI can assist by:
- enforcing specificity,
- flagging ambiguity,
- aligning grounds with your chronology.
Ground 2: identifying the legal duty
This is where many claims collapse.
Judicial Review requires:
- a legal duty,
- not just a power,
- and not just an expectation.
The question is:
Was the public body required by law to act — and did it fail to do so lawfully?
Without a duty, there is no unlawfulness.
AI can help:
- check whether you are assuming a duty,
- flag where a duty needs to be evidenced,
- prevent overstatement.
But you must verify the law.
Ground 3: choosing the correct public-law ground
Most JR claims rely on one (sometimes two) grounds:
Illegality
The decision-maker:
- misunderstood the law,
- failed to exercise a required power,
- or acted outside jurisdiction.
Procedural unfairness
The process was unfair because:
- no reasons were given where required,
- no opportunity to be heard was provided,
- mandatory procedure was not followed.
Irrationality
A very high threshold — rarely appropriate for litigants in person.
AI can help prevent the common mistake of:
- pleading all grounds “just in case”.
Courts view that as lack of focus.
The single biggest mistake: merits drift
Merits drift occurs when:
- arguments about fairness,
- disagreement with reasoning,
- or dissatisfaction with outcomes
creep into what should be a process challenge.
Examples of merits drift:
- arguing evidence should have been weighed differently,
- asserting bias without procedural basis,
- challenging findings of fact.
These are appeal issues — not Judicial Review issues.
AI is particularly useful here:
- it can flag evaluative language,
- identify opinion-based phrasing,
- and force re-framing into procedural terms.
Keeping law and fact separate (critical discipline)
Judicial Review requires:
- facts to be stated neutrally,
- law to be applied to those facts,
- not blended together.
A common error is embedding argument into factual narrative.
AI can help by:
- separating factual chronology from legal analysis,
- highlighting where language crosses the line,
- enforcing neutral drafting.
This separation builds judicial trust.
Alternative remedy: the silent killer of JR claims
Even where unlawfulness exists, Judicial Review may still fail if:
- an appeal route exists,
- or another adequate remedy is available.
Courts are firm on this.
You must:
- identify the appeal route,
- explain whether it exists in reality,
- and justify why JR is still appropriate.
This is where litigants in person often underestimate the burden.
AI can help:
- structure this justification,
- but cannot invent a lack of remedy where one exists.
Remedy: what you can (and cannot) ask for
Judicial Review remedies are limited.
You may ask for:
- a decision to be quashed,
- a matter to be reconsidered lawfully,
- a duty to be performed.
You cannot ask the High Court to:
- decide the underlying appeal,
- substitute its own view of the facts,
- grant compensation (save in rare cases).
AI can help test whether the remedy sought aligns with JR principles.
How AI should be used at the permission stage
AI is best used as a quality-control tool, not a generator.
Proper uses include:
- checking internal consistency,
- identifying merits drift,
- ensuring each ground maps to evidence,
- testing whether each ground answers the “so what?” question.
AI should not:
- expand arguments,
- multiply grounds,
- add speculative claims,
- generate case law without verification.
Permission-stage discipline is about less, not more.
The court’s perspective: what judges scan for first
Judges reviewing permission applications often:
- skim first,
- assess focus,
- test plausibility quickly.
They are alert to:
- scattergun pleading,
- emotional language,
- disproportionate claims.
A tight, restrained set of grounds signals seriousness.
Key Takeaways (for litigants in person)
- The permission stage is the real test in Judicial Review.
- Grounds must challenge lawfulness, not outcomes.
- Identify a legal duty — or the claim fails.
- Merits drift is the most common fatal error.
- AI is most useful as a:
- discipline tool,
- clarity enforcer,
- consistency checker.
- Fewer, stronger grounds beat many weak ones.
If you cannot state your grounds in calm, procedural language, Judicial Review is unlikely to succeed.
Preparing for the final stages
If permission is granted, the case moves into:
- full pleadings,
- possible disclosure,
- and substantive hearing.
But many litigants will face:
- permission refusal,
- or a conditional grant.
The final article in this series addresses that moment — and how to respond rationally.
Call to Action
If you are:
- preparing Judicial Review grounds,
- unsure whether your case has drifted into merits,
- or worried about permission-stage refusal,
You may wish to seek structured support before issuing proceedings.
Regulatory & Editorial Notice (JSH Law)
This article is provided for general information only and does not constitute legal advice.
Judicial Review claims are subject to strict procedural requirements and judicial discretion.
Improperly framed grounds may result in refusal of permission and adverse costs consequences.
Readers should seek independent legal advice where appropriate.
