AI is everywhere in 2026. Every software product claims to use it. But in legal claims work, where the difference between a correct case law citation and an invented one can mean the difference between winning and losing a dispute, "we use AI" is not enough. The question that matters is: what is the AI actually built on?
1. What a Language Model Does Well
Large language models are remarkable at understanding and generating human language. They can read a complex insurer letter and identify every argument, draft a response that reads naturally, and summarise lengthy judgments into plain language.
What they cannot do, on their own, is guarantee accuracy.
2. The Hallucination Risk in Legal Work
"Hallucination" refers to AI generating information that sounds authoritative but has no factual basis. In a legal context, this looks like a citation to a case that does not exist, a summary of a judgment that misrepresents the actual findings, or a principle attributed to a court that never stated it.
The danger is not that these errors are obvious. The danger is that they are convincing. A hallucinated case citation will follow the correct format, use realistic naming conventions, and be presented with the same confidence as a genuine reference.
For credit hire specifically, where the legal framework rests on a relatively small number of key authorities (Dimond, Lagden, Stevens, Pattni, McBride, Bunting), getting a citation wrong undermines credibility entirely.
3. Why a Knowledge Base Changes the Equation
A curated legal knowledge base is fundamentally different from the general training data of a language model. It is a structured, verified collection of specific authorities with confirmed citations, key principles, and judicial reasoning.
- Every citation can be traced to a verified source. The system retrieves rather than generates references.
- Principles are grounded in actual judicial reasoning. Not plausible-sounding language patterns.
- The scope of knowledge is defined and honest. A good system knows what it covers and what it doesn't.
4. The Three-Layer Problem
- Knowledge. Every case must be verified by someone who understands the law. Not automatable.
- Retrieval. The system needs to find the right authority for the actual question — not just anything that seems related.
- Output validation. Automated checks that catch errors before they reach the user.
5. What This Means in Practice
With a knowledge base: you paste an insurer's letter raising a BHR challenge citing Bunting v Zurich. The system retrieves verified BHR authorities and drafts a response that correctly addresses what Bunting actually decided. Every cited case exists and is applied to the right issue.
Without a knowledge base: the model drafts a response that reads well and sounds authoritative. But the case it cites may not exist — or may be cited for a principle it does not establish. You would not know without checking every citation yourself, which defeats the purpose.
6. The Trust Question
Trust in legal AI means specific things: every citation has been verified against the actual judgment; the system tells you when a question is out of scope rather than guessing; and the output is a starting point for professional review, not a finished product that bypasses human judgment.
A language model alone cannot deliver that. A language model combined with a curated knowledge base and proper validation layers can.
Disclaimer: this article is general guidance, not legal advice.
Frequently Asked Questions
- What is an AI hallucination in a legal context?
- It's when an AI generates a confident, well-formatted output — such as a case citation or judicial principle — that has no basis in reality. In legal work this is high-risk because the errors look authoritative.
- How does a curated knowledge base prevent hallucinations?
- The AI retrieves answers from a verified library of authorities rather than generating them from pattern recognition, so every citation is traceable to a real source.
Related articles
Basic Hire Rates (BHR): Why Bunting Doesn't Stop You Challenging the Facts
Bunting v Zurich is often cited to shut down BHR challenges. It doesn't. Here's what it actually decided and why rigorous factual challenges at first instance remain essential.
Impecuniosity in Credit Hire: What the Courts Actually Require
What does impecuniosity actually mean in credit hire? A practical guide to pleading, proving, and defending it, grounded in Lagden and recent case law.
When Insurers Try to Intervene: What the Case Law Actually Says
A strategy guide to defeating intervention arguments in credit hire, grounded in Copley v Lawn, Manton Hire, Sayce v TNT and Bee v Jenson.