Archive for month: January, 2024

The facts

In F Harber v HMRC [2023] UKFTT 1007 (TC) (4 December 2023) the Appellant, Mrs Harber, disposed of a property and failed to notify her liability to capital gains tax (“CGT”). HMRC issued her with a “failure to notify” penalty of £3,265.11. Mrs Harber appealed the penalty on the basis that she had a ‘reasonable excuse,’ because of her mental health condition and/or because it was reasonable for her to be ignorant of the law.

In a written document (“the Response”) Mrs Harber provided the Tribunal with the names, dates and summaries of nine First-tier Tribunal (“FTT”) decisions in which the appellant had been successful in showing that a reasonable excuse existed.  Remarkably, however, none of those authorities were genuine; they had instead been generated by artificial intelligence (“AI”).

The First-tier Tax Tribunal (“FTT”) accepted that Mrs Harber had been unaware that the AI cases were not genuine and that she did not know how to check their validity by using the FTT website or other legal websites. The FTT found she did not have a reasonable excuse, dismissed her appeal and upheld the penalty.  The FTT specifically stated that in coming to that decision, the Tribunal did not take into account her reliance on the AI cases

The Tribunal said:

‘In other words, our decision would have been the same if Mrs Harber had not provided the cases in the Response. Nevertheless, providing authorities which are not genuine and asking a court or tribunal to rely on them is a serious and important issue.’

 

The purported cases

The Appellant produced no fewer than nine cases in support of her submissions of ‘reasonable excuse’.   These cases were divided into two categories, those which related to “ignorance of the law” and those which related to mental health conditions.

The FTT said:

‘At the reconvened hearing, Mrs Harber said that the cases in the Response had been provided to her by “a friend in a solicitor’s office” whom she had asked to assist with her appeal. Mrs Harber did not have more details of the cases, in particular, she did not have the full text of the judgments or any FTT reference numbers.

Ms Man (HMRC’s representative) told the Tribunal that she had checked each of the cases in the Response to the FTT website, using not only the appellants’ names and the year as provided by Mrs Harber, but where the name was relatively common, she had extended the search to several years on either side. For example, when looking for “Smith v HMRC (2021)”, she had looked at cases between 2019 and 2023 where the appellant was called Smith. Despite that extended search, Mrs Man had not identified any FTT decision which matched the cases in the Response……

In considering whether the cases in the Response were genuine FTT judgments or whether they had been generated by an AI system such as ChatGPT, the Tribunal first carried out a review of other published judgments, and having done so, took into account the following points:

(1) None of the cases in the Response is included in the FTT website or other legal websites.

(2) Mrs Harber accepted that it was “possible” that the cases in the Response had been generated by an AI system, and she had no alternative explanation for the fact that no copy of any of those cases could be located on any publicly available database of FTT judgments.

(3) The Solicitors’ Regulation Authority (“SRA”) recently said[1] this about results obtained from AI systems:

“All computers can make mistakes. AI language models such as ChatGPT, however, can be more prone to this. That is because they work by anticipating the text that should follow the input they are given, but do not have a concept of ‘reality’. The result is known as ‘hallucination’, where a system produces highly plausible but incorrect results…..”

(4) The cases in the Response were “plausible but incorrect……

(a) The leading authority on the approach the FTT should take in reasonable excuse appeals is the UT judgment in Christine Perrin, commonly referred to simply as Perrin. The cited case of “David Perrin” uses the same surname and also concerns an appeal against a penalty on the grounds of reasonable excuse. However:

(i) the appellants have different first names;

(ii) the dates of the judgments are not the same; and

(iii) Christine Perrin lost her appeal whereas “David Perrin” succeeded.

(b) In the cited case of “Baker v HMRC (2020)”, the appellant challenged a penalty on the basis that his mental health difficulties provided him a reasonable excuse. This mirrors what happened in the Richard Baker judgment identified by Ms Man, see §17(1) above. However, that case was decided in a different year from the cited case, and Mr Richard Baker lost his appeal, unlike the appellant in the cited case.

(c) In the cited case of “Smith v HMRC (2021)”, the appellant successfully claimed a reasonable excuse on the basis of mental health difficulties. In Smith v HMRC [2018] UKFTT (TC) in which Mr Colin Smith similarly submitted that he had a reasonable excuse on the basis of “confusion and poor health”, but that case was again decided in a different year from the cited case, and Mr Colin Smith lost his appeal, unlike the appellant in the cited case.

(d) The FTT has decided 16 other reasonable excuse penalty cases in which the appellant’s surname was “Smith”, but none was issued in 2021, the year of the judgment cited by Mrs Harber, and none referred to mental health difficulties.

(e) The case of McMullen Holdings v HMRC [2011] UKFTT 327 (TC). That appeal concerned a VAT penalty for late registration and the FTT allowed the appeal. The cited case of “McMullen v HMRC (2018)” was similar in that the appellant successfully appealed a penalty, but that penalty had been charged for late filing of a tax return not for late registration for VAT, and the year was also different.

(f) The issue in Milner v HMRC [2014] UKFTT 735 (TC) was whether the appellant had a reasonable excuse for a VAT penalty; Mr Milner lost his appeal. The cited case of “Milner v HMRC (2020)” similarly concerned a reasonable excuse, but the penalty was charged for a different type of failure, in a different year, and the appellant won his appeal.

(g) The appellants in the other cases in the Response have the same surnames as those in reported decisions of other courts or tribunals. In BAILLI there are:

(i) five judgments in the last ten years in which one of the parties had the surname “Jewell”;

(ii) five judgments in the last ten years in which one of the parties had the surname “Oyesanya”;

(iii) twenty judgments in which one of the partes had the surname “Acheson”; and

(iv) four judgments in which one of the parties had the surname “Talal”.

(h) The wording of the cases in the Response is similar to that of published FTT decisions. To give just a few examples:

(i) The summary of “David Perrin” states that the appellant “argued that he was unaware of his obligation to notify HMRC and that the penalty charge was therefore unfair”. Numerous genuine FTT cases include as part of the published “key words” or headnotes, the phrase “appellant unaware of the obligation to notify… whether reasonable excuse”, see for example Clarke v HMRC [2020] UKFTT 144 (TC) and McDonough v HMRC [2020] UKFTT 421 (TC), in which the issue as to whether the penalty was “unfair” is also discussed.

(ii) The summary of “Jewell v HMRC (2016)” states that the appellant had submitted that “they had not been aware of the requirement to file a tax return as they had not received any correspondence from HMRC”. In the published case of One Motion Logistics v HMRC [2021] UKFTT 260 (TC), the appellant similarly claimed that “he had not been aware” of the relevant statutory requirement because the relevant correspondence “had not been received” from HMRC, and that as a result he had a reasonable excuse.

(iii) The summary of “Oyesanya” states that the appellant had submitted that “their mental health condition prevented them from being able to manage their affairs effectively”. In the published case of Freiberga v HMRC [2014] UKFTT 746 (TC), the appellant similarly claimed a reasonable excuse because she “was unable to manage her business affairs” as the result of “suffering from [an] acute bout of depression and suicidal thoughts”……

Having considered all the points set out above, we find as a fact that the cases in the Response are not genuine FTT judgments but have been generated by an AI system such as ChatGPT.

We also find as a fact that Mrs Harber was not aware that the cases in the Response were fabricated, and did not know how to locate or check case law authorities by using the FTT website, BAILLI or other legal websites.’

 

The Tribunal’s serious view of the fake cases

The Tribunal was clear that the production of cases which were not genuine was a matter of great concern.

‘Although we have accepted that Mrs Harber did not know the AI cases were not genuine, we reject her submission that this did not matter because the Tribunal had decided other reasonable excuse cases on the basis of ignorance of the law and/or mental health issues. We instead agree with Judge Kastel, who said on the first page of his judgment (where the term “opinion” is synonymous with “judgment”) that:

“Many harms flow from the submission of fake opinions. The opposing party wastes time and money in exposing the deception. The Court’s time is taken from other important endeavors. The client may be deprived of arguments based on authentic judicial precedents. There is potential harm to the reputation of judges and courts whose names are falsely invoked as authors of the bogus opinions and to the reputation of a party attributed with fictional conduct. It promotes cynicism about the legal profession and the…judicial system. And a future litigant may be tempted to defy a judicial ruling by disingenuously claiming doubt about its authenticity.”

The Tribunal then considered the substantive facts of the case and dismissed the appeal.

 

Conclusion

The uses of AI for legal purposes are very much at the forefront of our minds these days.  The perils of over-reliance on AI legal research, as per the above case, are not confined to England.

According to the Economist (6 June 2023)

‘No doubt many enjoyed a chuckle at the tale of Steven Schwartz, a personal-injury lawyer at the New York firm Levidow, Levidow & Oberman, who last month used Chatgpt to help him prepare a court filing. He relied a bit too heavily on the artificial-intelligence (ai) chatbot. It created a motion replete with made-up cases, rulings and quotes, which Mr Schwartz promptly filed after the bot assured him that the “cases I provided are real and can be found in reputable legal databases” (they were not, and cannot). Lesson learned, a tech-sceptic lawyer might conclude: the old ways are the best.

 Much has been made by legal writers about the potential application of AI.  For example, in a blog dated 6 January 2023, the  Allaboutlaw blog stated:

 ‘Legal research

 AI has transformed the way in which legal professionals conduct legal research, especially when searching for applicable case law. Large online legal data resources such as Westlaw and LexisNexis employ search algorithms that use AI. Legal professionals can provide the AI with relevant key terms and the algorithm will return tailored search results which only include applicable precedent.

 The tailoring of results means lawyers need not trawl through thousands of general and irrelevant search results, making legal research more efficient whilst limiting the time employees spend on repetitive work.

 Let us hope that the AI revolution in the legal research area turns out to be a boon, rather than a source of confusion.