In HMRC v Marc Gunnarsson [2025] UKUT 00247 (TCC), the Upper Tribunal (UT) agreed with HMRC that it was correct to recover coronavirus support payments made to Mr Gunnarsson under the self-employment income support scheme. Mr Gunnarsson was an “unrepresented litigant”, meaning that he represented himself before the UT.  

Mr Gunnarsson sent a draft skeleton argument to HMRC before the hearing. This referred to three decisions made by the First-tier Tribunal (FTT) which he said supported his case. Having been informed by HMRC that the decision did not exist, Mr Gunnarsson sent an amended and final skeleton argument that did not make reference to the decisions. During the hearing, Mr Gunnarsson accepted that he had used AI to help with his written submission.    

In commenting on Mr Gunnarsson’s use of AI, the UT refers to the case of Ayinde, R (On the Application Of) v Qatar National Bank QPSC & Anor [2025] EWHC 1383 (Admin), in which the Divisional Court takes a dim view of lawyers failing to check written legal arguments produced by AI, which it says runs the risk of false information being put before a court. In its decision, the Divisional Court sets out the current regulatory framework for lawyers and calls for more to be done to “address the misuse of artificial intelligence”.  

The UT makes it clear that legal representatives and unrepresented litigants are responsible for “the accuracy, both the reliability and credibility, of the information, both evidence and submissions, they present to the FTT or UT”. The decision ends with a warning that the UT takes this matter “very seriously” and may consider imposing sanctions on a party or representative found to have misused AI.  

Thankfully for Gunnarsson, the UT did not find him to be “highly culpable” as, not being appropriately trained or qualified, he “may not have understood that the information and submissions presented were not simply unreliable but fictitious”. However, parties to tax cases in the future may not be treated as leniently.  

This is not the first case involving an AI response that referred to fictitious sources. Other examples include the decisions of the FTT in Felicity Harber v HMRC [2023] UKFTT 1007 (TC) and Bodrul Zzaman v HMRC [2025] UKFTT 539 (TC).  

ICAEW has published a wide range of guidance and resources on AI, including how to manage the risks of using AI.