Using AI in the personal injury firm: How to avoid improper use

When I first started exploring artificial intelligence, I did what most people did. I opened ChatGPT.

Like many lawyers, I was curious. Could it summarise a document? Tidy up a letter? Help structure an argument? On the surface, it was impressive. Fast, confident, articulate. It felt like a glimpse into the future of legal drafting.

But it did not take long to realise something important.

Impressive is not the same as reliable, and fast is not the same as safe. In law, those distinctions matter.

Since joining LEAP as Head of Personal Injury Law, my focus has shifted from mainstream AI tools to specialist legal AI systems tailored to the realities of legal practice. That is where Matter AI and LawY came into view.

And that is where the conversation changes.

The risk of AI lies in improper use

The recent Upper Tribunal warning, reported in Legal Futures (24 February 2026), could not be clearer. A solicitor admitted to uploading client emails and Home Office decision letters into ChatGPT to improve and summarise them. The Tribunal described this as placing confidential information “in the public domain”, breaching client confidentiality and waiving legal privilege.

That is not a minor procedural issue. That is a regulatory issue.

At the same time, we have seen cases such as Ayinde and Al-Haroun, where fictitious case citations — so-called “hallucinations” — found their way into court documents. The court’s response was firm: responsibility remains with the lawyer. Whether the error was made by a trainee or by AI is irrelevant, as the solicitor on record carries the duty.

As the legal industry continues evolving and technology advances, AI is not going away. Anyone with access to a search engine has access to AI, so it would be unrealistic for firm managers to ban its use entirely. So the real question is not whether lawyers will use AI, but whether they will use the right kind.

From OpenAI curiosity to legal AI discipline

My own journey mirrors what many practitioners are experiencing. ChatGPT was a useful introduction, as it demonstrated potential. But it also exposed limitations:

• It does not understand your matter.

• It does not operate inside your case management system.

• It is not designed around legal compliance.

• It can fabricate authorities.

• And unless carefully configured, it is not a closed legal environment.

That led me to explore AI built for lawyers, inside legal workflows, with professional safeguards embedded from the start.

Two stand out: Matter AI and LawY.

Matter AI

Matter AI is not a general chatbot bolted onto legal work. It sits within your LEAP matter and securely works with your data.

It searches your matter, not the internet

Matter AI intelligently scans your documents, emails, attachments, scanned PDFs and even handwritten notes using OCR. Whether there are 10 documents or 10,000, it can locate key dates, contact details, appointments, inconsistencies and insights in seconds — and reference the source files.

That eliminates the time-consuming manual trawl that eats into profitability.

It allows inline editing in Word

You can draft using structured AI prompts directly in Microsoft Word, then edit the output inline. Expand reasoning. Adjust tone. Remove irrelevant sections. You stay in control of the drafting process.

AI becomes an assistant, not the author of record.

It offers built-in legal prompts

With over 200 AI prompts covering affidavits, statutory declarations, briefs to counsel, and more, it guides on structure, annexures, formatting, and compliance considerations. That is a practical productivity gain.

It ensures secure drafting

Draft emails can be generated from matter data and sent directly to Outlook with recipient and subject fields populated. Again, this happens within your firm’s controlled system, and not by uploading confidential material to an open platform.

Matter AI focuses on operational efficiency within a secure legal environment. It reduces friction. It does not compromise confidentiality.

LawY

Where Matter AI works within your matter, LawY addresses legal research and drafting, providing AI-generated answers to legal questions. Those answers can then be submitted for verification by qualified UK Solicitors before being returned as verified responses, a two-step process that is crucial for lawyers.

In light of Ayinde and Al-Haroun, the Divisional Court made it abundantly clear: lawyers must check AI-generated research against authoritative sources, so LawY is built with that obligation in mind.

It assists with:

• Legal research

• Drafting letters and court documents

• Preparing affidavits

• Reviewing case law and legislation

• Proofreading and summarising

And crucially, it links back to authoritative materials, making verification easier rather than leaving the burden entirely on the user.

LawY does not replace professional judgment. It accelerates the groundwork while keeping accuracy and accountability front and centre.

For firms concerned about hallucinations, compliance and regulatory exposure, that distinction is fundamental.

Why the distinction between open and closed AI matters

Using an open AI platform for generic brainstorming is one thing. Uploading client documentation or relying on it for case citations is quite another. As my colleague Catherine Gaynor rightly highlighted, the problem is not AI itself, but the uncritical reliance on the wrong tool.

Open Systems (e.g. generic GPT tools)

• May use prompts to further train models

• Exist outside your case management system

• Can expose confidential information if misused

• Are not designed around legal regulatory frameworks

• Are prone to hallucination if used for research

Closed Legal Systems (e.g. Matter AI within LEAP)

• Operate within your firm’s secure environment

• Do not place client information into the public domain

• Are built specifically for legal workflows

• Integrate with Matter Data

• Support compliance rather than undermine it

The key lesson I took away from the recent cases is that the court does not care whether the error was made by a trainee or by artificial intelligence; the solicitor remains responsible.

AI adoption can be transformative as long as it’s done right

We are at a turning point, where firms can either ban AI and pretend it will disappear (it will not), or lead its responsible adoption with training, policy, and purpose-built tools.

From my perspective, the path forward is clear. Firms should train their legal teams to use AI as a tool, not as an authority. Even more importantly, they should move away from generic platforms built for mass use, and instead use systems designed for legal practice, embedding verification, supervision and professional judgment into every workflow.

The future of legal practice will include AI. The only real question is whether firms adopt it carelessly or strategically.

Used properly, AI adoption can support sustainable workloads, improved profitability and better client outcomes without compromising confidentiality or compliance.

About the author

With extensive experience in Personal Injury litigation, William O’Brien previously served as Partner at Express Solicitors, where he led teams managing catastrophic injury/serious injury/multi-track and fast-track injury claims.

As Head of Personal Injury Law at LEAP Legal Software, William ensures that technology is designed to meet the real needs of PI practitioners, enhancing compliance and client care through innovation.