Since Chat GPT burst onto the scene and heralded the arrival of AI into ‘normal’ personal and professional lives there has been lively and anxious debate as to whether it is friend or foe.
In the case of R (Ayinde) v London Borough of Haringey [2025] EWHC 1040 (Admin) it was exposed as a foe, at least to one pupil barrister (https://caselaw.nationalarchives.gov.uk/ewhc/admin/2025/1040?query=ayinde).
Counsel (a second six pupil) was instructed by a law centre to settle the grounds of judicial review in a housing matter. The grounds relied upon a number of authorities (cases) to establish relevant points of law that supported the grounds.
It was pointed out by the solicitors on the other side (LB Haringey) that some of those cases did not exist. The law centre provided an unsatisfactory response which did not provide an explanation for what they described as ‘citation errors’. Joshua Rozenburg has reported that it has subsequently revealed in another hearing that the response was drafted by counsel (https://rozenberg.substack.com).
LB Haringey made an application for wasted costs – that is an order that the law centre and counsel pay the local authority’s costs on the basis that they have acted in a way that is improper, unreasonable or negligent.
At the hearing of the wasted costs application, before Mr Justice Ritchie, counsel gave an explanation which was found to be itself untrue
Mr Justice Ritchie was extremely critical of the inclusion of five ‘fake cases’ in the grounds. He considered that it was a breach of professional standards as was the fact that in court counsel described the false cases as ‘minor citation errors’. The solicitors were also responsible for ensuring the accuracy of the grounds and should have been shocked when they were told by the other side the cases did not exist.
During the hearing it was suggested on behalf of Haringey that the cases may have been obtained via AI. Mr Justice Ritchie was not able to make a finding on that but was of the view that using cases obtained via research using AI without checking them would be negligent.
A wasted costs order was made and it was directed that a transcript of the judgment be sent to the Bar Standards Board and Solicitors Regulation Authority.
Subsequently, the case was back in court last week to consider what further steps the court should take against counsel, paralegal and the law centre, including whether they should face proceedings for contempt of court.
The Law Society Gazette has reported on that hearing before the President of the King’s Bench Division, Dame Victoria Sharp, which has not yet reached a conclusion (https://www.lawgazette.co.uk/news/barrister-did-not-appreciate-seriousness-of-fake-citations-high-court-hears/5123387.article?utm_source).
My understanding of the Gazette report is that at this hearing it was acknowledged that the cases were or may have been AI generated but that had not been appreciated by counsel at the time. She had not used Chat GPT but it appears she accepted that other internet searches returned AI generated results. It was explained on her behalf that she had little by the way of resources provided by her chambers and undertook research online, creating a table of cases that she worked from. She had since deleted this table on advice from a senior member of chambers. She did appreciate the seriousness of the fake citations and felt that there would be alternative authorities to support the propositions put forward.
This is a quite a different explanation to that which was provided to Mr Justice Ritchie at the earlier hearing in which it was said that she had photocopied at least one of the cases. We shall have to wait and see what conclusions are reached by the court on that when judgment handed down.
I think that it is fair to say that there is sympathy within the legal community for a pupil barrister who was so under-resourced and apparently unsupported by her chambers that she did not have access to the White Book or online legal resources. However, as many have pointed out the Inns of Court have excellent libraries (all barristers are members of an Inn and therefore have access to their resources).
Online research is now the norm but many of us will recall when that was not the case and we undertook the majority (if not all) of our research using hard copy books and reports. Of course it takes longer and is less convenient but accuracy is non-negotiable when you are presenting documents to the court.
It is only possible end up in a position of presenting non-existent cases to the court if you have not looked up and read the judgment. If you don’t look up the case reports then you are relying, on the online article or AI not only to deliver to you a case that actually exists, but also on their summary of the ratio decidendi (the authoritative point of the case). Even if you are using a professional text book it is not wise to cite a case that you have not read. In doing so you are fundamentally limiting your understanding of the authority.
Whether there was a lack of supervision and training here or something else contributing to the events, there can be no doubt that Ayinde will send a shiver down the spine of most legal professionals.
I confess that I have approached the possibility of using AI with some considerable caution (you can read that as – I have never even opened Chat GPT). One of the reasons for my restraint is that my instincts are to always know where information or evidence is coming from. I think that most, if not all, lawyers will feel the same even if many are more open to the prospect of using AI than I have been so far.
When undertaking research I ask myself who is telling me this and what might be their motivations for doing so? Do they have a vested interest in this topic or my decision? Do they have the skills and knowledge to give a reliable and complete answer to this question? What are their political views and are they relevant here? Is there any other source that supports this one?
These elements of critical thinking are just as important in personal life when, for example, reading the news or researching the best nursery for my children to go to.
So, the prospect of using AI to gather an answer from the internet without being able to closely analyse its sources is unattractive, even in a non-professional context.
I am certain that there are important uses for AI in all aspects of our lives. But it is clearly incumbent on us, as professionals, to understand the technology that we are using and to have a full appreciation of its risks as well as its benefits. That must include being aware of when AI is a part of any other technology that we are using so that we know when we are encountering it and act accordingly.
I would be surprised if there are any of us that ought to be using AI in our practice without first undertaking appropriate training and giving careful consideration to the Code of Conduct. That will certainly be my starting point.
There is guidance available from the Bar Council on the use of generative AI (https://www.barcouncil.org.uk/resource/new-guidance-on-generative-ai-for-the-bar.html).
There is also judicial guidance which is of interest (https://www.judiciary.uk/guidance-and-resources/artificial-intelligence-ai-judicial-guidance-2/).