Written by AI can help a Litigant in Person organise a case, draft clearer documents and test arguments. But when it is treated as a lawyer, judge or legal database, it can become a fast route to false confidence, fabricated authorities and procedural damage.
Publication snapshot
- The article explains why AI can assist legal preparation but cannot replace legal judgment.
- It identifies four common failure points: false case law, procedural errors, confirmation bias and overconfidence.
- It gives LiPs a practical verification protocol before using AI-generated material in correspondence, applications or submissions.
- It reinforces the core rule: AI may help draft the work, but the litigant remains responsible for what is filed.
The rise of AI in legal work
Artificial intelligence is now part of the legal landscape. It can summarise documents, help draft correspondence, generate chronologies, suggest issue lists and support legal research. For Litigants in Person, that accessibility is attractive. It appears to offer structure, speed and confidence in a system that often feels technical and hostile.
Used properly, AI can be useful. It can help a LiP turn a chaotic bundle into a working chronology. It can help shorten an emotional email. It can suggest questions to ask a solicitor, identify gaps in evidence, or produce a first draft of a witness statement structure.
The problem begins when the LiP treats AI output as verified legal work. AI does not know the case. It does not owe duties to the court. It does not understand the procedural context unless the user provides it accurately. It can sound confident while being wrong.
When AI misleads instead of assists
Consider the common LiP scenario. A self-represented party believes AI can help level the playing field against a legally represented opponent. The LiP asks for authorities, submissions and procedural arguments. The answers look polished. The citations appear plausible. The draft sounds more legal than anything the LiP could have written alone.
The danger is that polish can be mistaken for accuracy. In one case-study pattern, the LiP relied heavily on AI-generated legal research and drafting. The result was not empowerment, but avoidable damage: misapplied case law, incorrect procedural references and arguments that reinforced the LiP’s preferred view rather than testing the weaknesses of the case.
The LiP asks for support
AI is asked to find cases, draft arguments and identify procedural routes.
The output looks persuasive
The response uses confident language, legal vocabulary and apparent authorities.
The material is filed
The LiP assumes the answer is accurate and does not verify every case, rule and quotation.
The damage appears
The opponent exposes the error, credibility falls and the court becomes less receptive.
This is the central lesson. AI can make a weak argument look coherent. It cannot make a legally unsound argument good.
Four litigation risks
The risks of AI use in litigation are not theoretical. They are practical, immediate and capable of affecting how a court or tribunal sees the litigant.
Misapplied case law
AI may identify a real case but misunderstand what it decided. A case that appears supportive may actually undermine the argument once read properly.
Procedural inaccuracy
AI may cite the wrong rule, an outdated rule, the wrong forum or a procedural step that does not apply to the hearing in question.
Confirmation bias
If the user asks leading questions, AI may return answers that reinforce the user’s theory rather than testing the opposing argument.
Excessive litigation
AI can create false confidence, encouraging applications or appeals that lack legal merit and may be viewed as unreasonable.
The most dangerous AI error is not always the obvious hallucination. Sometimes the case exists, the rule exists and the wording sounds right — but the legal context is wrong. That is harder for a LiP to spot.
AI can assist the preparation of legal work. It cannot relieve the person filing that work from responsibility for accuracy, relevance and judgment.
The verification protocol
Before using AI-generated material in legal correspondence, applications, witness statements or submissions, a LiP should run a basic verification protocol. This is not optional. It is the difference between using AI responsibly and outsourcing judgment to a machine.
Before filing AI-assisted material
- Check every case citation: confirm the case exists on an official or reliable legal database.
- Read the relevant passage: do not rely on an AI summary of what the case supposedly says.
- Check the procedural rule: confirm it applies to the correct court, tribunal, jurisdiction and date.
- Test the opposite argument: ask what the strongest response from the other side would be.
- Remove overstatement: replace dramatic conclusions with precise legal propositions.
- Seek review where possible: get a solicitor, adviser or experienced supporter to check important filings.
A simple rule helps: if the LiP cannot personally explain what a case proves, why it applies and where the proposition appears in the judgment, it should not be relied on.
How LiPs should use AI
The answer is not to avoid AI altogether. The answer is to use it for the right tasks and refuse to use it for the wrong ones.
Safer uses
- Creating a first draft structure.
- Summarising a party’s own documents.
- Turning a long narrative into a chronology.
- Preparing questions for legal advice.
- Reducing emotional wording in correspondence.
High-risk uses
- Finding authorities without later verification.
- Drafting submissions on unfamiliar law.
- Interpreting procedural rules without checking them.
- Assessing prospects without independent review.
- Generating allegations, fraud claims or misconduct complaints.
AI is strongest when the LiP uses it to organise material the LiP already has. It is weakest when asked to invent the law, predict judicial behaviour or validate a preferred theory.
Use AI to structure the problem.
Verify law and procedure independently.
Cut anything that cannot be checked.
File only what you can defend.
Closing point
AI is reshaping legal preparation. For LiPs, that can be positive. It can make documents clearer, reduce drafting time and help identify the real issues in a case.
But AI is not a lawyer. It does not know when it is out of date. It does not know when it has invented a case. It does not know whether a rule applies to a particular tribunal. It does not protect a litigant from the consequences of filing inaccurate material.
The safe approach is disciplined and simple: use AI to assist, not decide. Verify everything. Think like a litigant who remains responsible for every word placed before the court.
Reference points
Courts and Tribunals Judiciary, Artificial Intelligence (AI) Judicial Guidance.
Ayinde v London Borough of Haringey / Al-Haroun v Qatar National Bank [2025] EWHC 1383 (Admin).
Harber v HMRC, First-tier Tribunal tax appeal involving AI-hallucinated authorities.
Disclaimer
This article provides general strategic commentary for informational purposes only. It does not constitute legal advice. AI tools vary, legal procedure is forum-specific, and every case turns on its own facts, rules and evidence. Litigants should verify all legal authorities and procedural rules from reliable sources and seek advice from a suitably qualified solicitor or regulated adviser where necessary.
In a case brought by a ground floor tenant and a tenant of a first floor flat (same building) both held on a long lease of residential property did under the LRHUD Act 1993 who by its their own admission without the landlords/owners permission or consent, both acting in concert acted in breach of the terms expressed in the Lease, and did obtain the services of an unlicenced contractor to remove and replace a party wall containing Asbestos, contrary to the Asbestos Regulations 2012 rendering the breach irremediable and the property uninhabitable, uninsurable, and unsaleable and further in breach of the Fire Safety Order 2005. The Claimants obtained a signature from the District Judge granting a new lease notwithstanding that the Application Notice itself under the 1993 Act concealed their breaches by non disclosure of lethal defects both material threats to life which the District Judge sitting at the CLCC granted the new lease. so —Thank THE LORD FOR AI AND CHAT GPT likely to bring an end to corruption.