Since our establishment in 1783, Nalders has practised litigation in various capacities. We have acted in cases against parties represented by leading law firms, as well as those acting as litigants in person.
Litigation is expensive. That reality is well understood by those working in the field, and it is certainly not lost on those on the receiving end of a claim.
Understandably, some individuals attempt to defend claims themselves. Some even bring claims without legal representation. In many cases, that approach can work. The small claims track of the County Court is designed to support access to justice without lawyers.
However, when claims are large, legally complex or procedurally demanding, proper advice is not simply helpful. It is often essential.
Free or low-cost fixed fee advice does exist, although it can be harder to find in Cornwall. Despite this, the firm has seen an alarming increase in litigants in person using generative AI as a substitute for legal advice, often to their serious detriment.
The documents produced are almost always the same: confidently written, superficially polished, and fundamentally flawed.
Not every litigant in person uses AI. However, those of us who practise litigation can often recognise the signs immediately. In some cases, the defects in these documents have caused real and lasting procedural damage. In others, they have exposed litigants to risks they plainly did not understand when signing and filing documents.
We have also seen substantial adverse costs orders made, not to penalise parties for using AI, but because AI has led them to pursue positions or arguments that were not legally sustainable. In our view, many of those costs consequences could have been avoided.
This article is not intended as criticism. It is intended as a warning. The danger lies not in the intelligence of litigants in person, but in the false confidence that AI can create.
AI Does Not Understand the Law, the Case, or the Consequences
This article focuses on one of the most important documents in any claim or defence: the statement of case. This includes particulars of claim, defences, and counterclaims.
Under the Civil Procedure Rules (CPR), statements of case must comply with CPR Part 16 and, where applicable, Practice Direction 16. They must set out a concise statement of the facts relied upon, identify the legal basis of the claim or defence, and plead only what is properly arguable.
AI does not analyse evidence, assess costs risk, or understand procedural requirements. It generates text based on the information it is given, producing output that may sound plausible but is not necessarily accurate.
In statements of case reviewed by the firm where AI appears to have played a significant role, we have seen non-existent causes of action, inconsistent or incoherent pleadings, defective statements of truth, and extensive factual assertions unsupported by evidence.
To a trained lawyer, such defects are often obvious. For a litigant in person, however, they can be much harder to detect, particularly when the document is written with confidence and apparent authority, and appears to fully support their position.
The Statement of Truth: Not a Technicality
Statements of case must be verified by a statement of truth in accordance with CPR Part 22 and Practice Direction 22. Compliance is mandatory.
One of the most recurring and basic errors we see concerns the statement of truth. A defective statement of truth is not a minor procedural misstep.
When a litigant signs a statement of truth, they are certifying the truth of the contents. They are also confirming that they understand proceedings for contempt of court may follow if false statements are made without an honest belief in their truth.
For particulars of claim signed by a claimant, the correct form is:
“I believe that the facts stated in these particulars of claim are true. I understand that proceedings for contempt of court may be brought against anyone who makes, or causes to be made, a false statement in a document verified by a statement of truth without an honest belief in its truth.”
In AI drafted documents we frequently see instead:
“I believe that the facts stated in this document are true.”
This is not compliant. A defective statement of truth can lead to the document being rejected, struck out, or relied upon less readily by the court. It can also expose a litigant to serious costs consequences.
When a document is filed at court, the litigant is responsible for its accuracy and compliance with the rules. Being a litigant in person does not dilute those obligations, and responsibility cannot be delegated to software.
AI Does Not Advise on Risk
Costs risk is governed principally by CPR Parts 44 to 47. The court has wide discretion as to costs, including whether costs are proportionate, reasonably incurred, and whether a party has acted reasonably in the conduct of proceedings.
AI tools cannot provide tailored advice on the consequences of getting litigation wrong. They do not have access to the full facts, evidence, and procedural context of the case, nor can they properly assess litigation risk.
In addition, the more documents that are uploaded to an AI system, particularly if unredacted, the more personal data is transferred to that service. This increases confidentiality and data protection risks for all parties.
Litigation itself is inherently risky. Costs continue to rise as a matter progresses, and ultimately someone must bear them. Cases involving litigants in person often cost the represented party more, largely because lawyers are required not to take advantage of unrepresented individuals.
AI can make this worse. It often produces lengthy and unfocused material which must be carefully reviewed. Litigants in person may become more entrenched as AI repeatedly echoes their position, particularly if they provide only their own version of events. As a result, they may resist reasonable discussion or explanations of procedure.
If they then lose and costs are awarded against them, the scale of those costs often comes as a shock.
Over-validation and Entrenchment
AI is designed to be engaging and affirming. It is built to encourage repeated use. In a legal context, this can result in systematic over-validation.
This firm has reviewed an exchange between a litigant in person and ChatGPT that was voluntarily disclosed. The pattern was clear. The AI reinforced the litigant’s preferred narrative and optimism as to their chances of succeeding, because it was only provided with their opinion of the case. It failed to identify fatal weaknesses, downplayed evidential and legal hurdles, and presented speculative arguments as though they were settled law.
For litigants in person, this can create a dangerous false sense of security. It can encourage the advancement of legally unsustainable arguments and, in some cases, positively harmful ones.
AI as a Tool, Not a Lawyer
This is not to say that litigants in person should never use AI. We would never seek to restrict the resources available to assist them. AI can be a helpful tool.
However, it should be used as a tool and not as a substitute for legal advice.
The firm uses AI for administrative and supportive tasks, such as directing research, spellchecking, and improving grammar in text we already understand. In fact, it was used to assist with research and proofreading of this article. It can also be useful for identifying topics for further research, both for lawyers and for litigants in person.
However, AI should not be used to draft court documents, to state the law, or to replace professional advice.
The use of AI in the legal industry is only set to increase, particularly with the development of paid products such as Lexis+ AI, CoCounsel (formerly Casetext), and Harvey. These products are trained differently from free, general AI tools such as ChatGPT and Copilot. However, they are still designed to assist lawyers, not to replace them.
It’s Not Just Litigants in Person
Recent court decisions have exposed how quickly AI misuse can compromise the accuracy of legal submissions, including in cases involving legal professionals.
One prominent example arose in an £89 million claim against Qatar National Bank. The claimants filed 45 supposed authorities, only for the court to discover that 18 did not exist at all. The solicitor with conduct of the matter admitted he had relied on legal research supplied by his client without verifying it.
A similar issue emerged in proceedings brought by Haringey Law Centre, where a pupil barrister cited five cases which were later revealed to be entirely fictional. The work was not checked by the supervising solicitors.
These incidents resulted in referrals to the relevant professional regulators. They also prompted the High Court to issue a detailed warning under the Hamid jurisdiction about the misuse of AI in litigation, emphasising the serious implications such conduct poses for the administration of justice.
Both the Solicitors Regulation Authority (SRA) and the Bar Standards Board (BSB) have cautioned practitioners against relying uncritically on AI generated material. They have stressed that legal professionals must thoroughly verify the accuracy of any AI output before placing it before a court.
The SRA has highlighted the risks of inaccuracy, bias, confidentiality breaches, and the ongoing accountability of solicitors for all submitted work, even where AI tools are used. Similarly, the BSB has warned that AI can only be safely integrated into legal practice when barristers understand its limitations and maintain proper oversight. Professional responsibility cannot be delegated to an algorithm.
In addition, the Law Society has issued guidance on the responsible use of generative AI, underscoring the need for vigilance, careful supervision, and ethical awareness as these technologies become more prevalent in legal work.
In our view, this guidance is just as important for litigants in person who choose to use AI as it is for legal professionals. Although litigants in person are not subject to the same regulatory obligations, they still engage with the court system and should act in a way that supports, rather than undermines, the proper administration of justice.
Conclusion
The court’s approach to litigants in person is well established. While allowances may be made for unfamiliarity with procedure, litigants in person are still required to comply with the Civil Procedure Rules and Practice Directions.
The use of AI does not lower that standard, nor does it transfer responsibility for compliance.
The firm has now seen enough AI generated pleadings from litigants in person to be confident of this conclusion: AI is not helping you.
In litigation, confidence is irrelevant without accuracy. If a litigant signs and files a document, the court will treat every word as theirs, regardless of who or what generated it.
If you would like advice on a contentious matter from a trained and experienced practitioner, please get in touch with our litigation team.
This article is for general information only and does not constitute legal or professional advice. Please note that the law may have changed since this article was published.