In partnership with
I anticipate this week's lead story will be circulating in most law firm inboxes before this newsletter reaches you, and it should be. It is a stark warning for us all. ICC Judge Mullen has published a 96-paragraph judgment after a Pinsent Masons solicitor used generative AI to fabricate a non-existent statutory provision in correspondence with the High Court. The firm has self-referred to the SRA, and the judgment includes the full AI chat transcripts as part of the public record.
AI in Practice
Pinsent Masons self-refers to SRA after AI fabricates insolvency rule in court correspondence
What should have been a routine insolvency block transfer application has become a cautionary tale for the profession. In Cork v Smith [2026] EWHC 1199 (Ch), ICC Judge Mullen examined how a junior solicitor at Pinsent Masons (referred to as "LA" in the judgment) used generative AI to draft two letters to the High Court, both of which contained misleading statements of the law.
The sequence of events is worth setting out. The judge queried whether the court had power to grant a particular form of release to a liquidator. LA turned to the firm's AI tool to research the question. The AI hallucinated a non-existent sub-rule (rule 12.37(5) of the Insolvency Rules 2016) and produced fabricated statutory wording, which LA included in a letter to the court without checking the actual legislation. When the judge identified the problem, LA used the AI again to construct an explanation for the first letter, producing a second misleading response.
The AI chat transcripts make for uncomfortable reading. They run to 59 pages and are now part of the public court record. They show the AI repeatedly warning LA to verify its output against authoritative sources. LA does not appear to have done so at any point. The judge noted that LA "seems to have almost entirely outsourced the thinking process" to the program. The supervising solicitor and reviewing partner both accepted they should have caught the errors but did not, and neither was aware that AI had been used in the drafting.
ICC Judge Mullen accepted there was no deliberate intent to mislead and concluded that public admonishment, combined with the firm's self-referral to the SRA, was a proportionate response. But the reasoning makes clear what was already suspected by this writer. That accountability for AI output sits with the solicitor, not the tool. Endorsing the framework set out by Dame Victoria Sharp P in R (Ayinde) v London Borough of Haringey [2025] EWHC 1383 (Admin), the judge confirmed that legal professionals cannot outsource legal research or reasoning to AI. It is, as he put it, "a tool to be used with caution."
Pinsent Masons initially self-referred the supervising solicitor and reviewing partner to the SRA. Following the hearing, the firm confirmed that LA would also be referred. All three are now under regulatory scrutiny.
Takeaways
Act: Review your firm's AI supervision arrangements now. Ensure that juniors understand they must verify AI-generated legal research against primary sources before it is used in any document, and particularly in court correspondence. If your AI policy does not require disclosure of AI use to supervisors, this judgment suggests it should.
Watch: The SRA investigation. The outcome will signal how the regulator treats AI-related professional conduct failures, and whether the existing standards and regulations are considered sufficient or whether specific AI guidance will follow.
Risk: The judgment confirms that AI chat transcripts can be ordered disclosed and published as part of the court record. Firms should consider whether their internal AI usage policies address the discoverability of AI interactions, and whether fee earners understand that their AI conversations may not remain private.
On your radar
ICO automated decision-making consultation closes tomorrow (29 May): The ICO's consultation on its draft guidance for automated decision-making, which includes profiling, closes at 23:59 on 29 May 2026. The guidance follows changes to the ADM regime introduced by the Data (Use and Access) Act 2025 and will shape how the ICO enforces ADM requirements under UK GDPR going forward. Why it matters for UK lawyers: Any firm using AI for triage, client intake, document classification, or risk scoring should consider whether those processes fall within the ADM definition. This is also relevant to firms advising clients who deploy AI in customer-facing decisions. Respond via Citizen Space or by emailing [email protected] before the deadline. (ICO)
UK AI hallucination cases in court reach 64: The Natural and Artificial Law tracker (a blog thoroughly recommended by the writer), which logs documented instances of AI-generated false authorities or fabricated legal material reaching court proceedings, now records 64 UK cases. Some of the most recent additions (Re A, B, C, D [2026] EWFC 71 and Brightwaters Energy v Eroton [2026] EWHC 296) were both identified by judges, not by the lawyers who submitted the material. Cork v Smith is the big news this week but far from an outlier. Why it matters for UK lawyers: 64 is the number that made it into the record. The actual figure is certainly higher, since cases where hallucinations are caught before filing leave no trace. If your firm does not have a mandatory verification step for AI-assisted legal research, the question is no longer whether a hallucination will reach a court but when. (Natural and Artificial Law)
Ad Break
In order to help cover the running costs of this newsletter, please check out the advert below. In line with my promises from the start, adverts will always be declared.
Turn AI into Your Income Engine
Ready to transform artificial intelligence from a buzzword into your personal revenue generator?
HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
For Review
"The use of artificial intelligence: landing solicitors in serious trouble again (and don't blame the AI for everything)" (Civil Litigation Brief)
A detailed walkthrough of the Cork v Smith judgment with a particular focus on one overlooked point: the AI itself repeatedly warned the solicitor to check its output against authoritative sources. The author argues that this makes it harder to frame the problem as an AI reliability issue and easier to frame it as a supervision and professional competence issue. Worth reading alongside the judgment itself for the practical lessons on where the failures actually occurred.
Read or listen: Civil Litigation Brief
"New guidance supports barristers to safely adopt Artificial Intelligence and emerging technologies" (Bar Standards Board)
The BSB published its formal AI guidance on 18 May, the first regulator-specific framework for the Bar. It treats responsible AI use as a competence and practice-management issue under existing Core Duties, rather than introducing new rules. Two points are worth noting. First, the BSB warns that free tools such as ChatGPT will "generally be unsuitable" for legal work because their terms typically allow providers to store prompts and use input data for training, which is difficult to reconcile with confidentiality obligations. Second, barristers are not automatically required to disclose AI use to clients or the court, but must be transparent where AI materially affects the nature or scope of the service. Read alongside Cork v Smith, where the court ordered the full AI chat transcripts disclosed, the gap between "no automatic duty to disclose" and "your transcripts may be ordered into the public record" is one practitioners should think about now.
Read: BSB Guidance (PDF) | BSB announcement | Legal Cheek
Practice Prompt
This week's prompt is for verifying AI-generated legal references, which feels particularly apt. Ensure you fill in the placeholders marked with {}. Remember to adhere to the Golden Rules and do not upload confidential or privileged information to public tools.
I have a draft {letter / skeleton argument / advice note / submission} which contains references to {statutes / statutory instruments / rules / case law}. The draft was {partially generated using AI / produced using AI-assisted research} and I need to verify every legal reference before it is sent.
For each legal reference in the text below, confirm:
1. Whether the provision or case cited actually exists
2. Whether any quoted or paraphrased wording accurately reflects the source
3. Whether the provision is in force and has not been amended or repealed
Flag anything that: (a) cites a provision that does not exist, (b) attributes wording to a source that does not contain it, (c) cites a provision that has been repealed or amended in a way that changes its effect, or (d) mischaracterises the effect of the provision. For each flagged item, explain the problem and provide the correct position.
Area of law: {e.g. insolvency, employment, commercial litigation}
Jurisdiction: {England and Wales / Scotland / Northern Ireland}
[Paste the draft text here]
How did we do?
Hit reply and tell me what you would like covered in future issues or any feedback. We read every email!
Thanks for reading,
Serhan, UK Legal AI Brief
Disclaimer
Guidance and news only. Not legal advice. Always use AI tools safely.
Recommended Newsletters
Below are a few newsletters that I recommend, for various reasons. Check them out!
