Erzatz not echt: AI and case law

To a large degree the First-tier Tribunal (‘FTT’) case of Felicity Harber v HMRC [2023] UKFTT 1007 (TC) is unexceptional.  The lady in question had failed to tell HMRC about a capital gain she had made on disposing of a property.  HMRC had imposed a penalty for her failure.  The lady had appealed claiming that she had a reasonable excuse for the failure, but the FTT had held that she had none.  Such cases are not unusual.

What was somewhat out of the ordinary was that Mrs Harber had provided the FTT with the names, dates and summaries of no less than nine previous FTT decisions supporting her case, which the FTT disregarded in coming to its decision.

What was truly extraordinary was the reason for that disregard.  Not a single one of the nine decisions was genuine.  They were all based on material which Mrs Harber had found on the internet and which had apparently been generated by Artificial Intelligence.

Obviously, citing made-up cases in court is a serious matter and not to be recommended.  In a US case a law firm was fined $5,000 after fake citations generated by ChatGPT were submitted in a court filing.  However, the FTT accepted that Mrs Harber was unaware of the unreliable provenance of the cases she had cited: she had simply not known how to check their validity using the FTT website or other legal websites.

Quoting from a recent report by the Solicitors Regulation Authority, the FTT observed that:

‘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.’

The narrow moral of the case: if you are going to represent yourself before the FTT, do check that you know what you are doing: if not, instruct someone who does.

The wider moral, and the more worrying one, is that with the increasing advance of artificial intelligence it is becoming increasingly difficult everywhere (not only in the legal world) to differentiate reality from Matrix-esque fantasy.  Keep your wits about you.

For more information, please get in touch with your usual (human) contact at BKL or use our enquiry form.

NICOLA HALL

BILSHAN MENSAH

Sam Inkersole

In 2022, Sam won the Taxation’s Rising Star award at the Taxation Awards in and was named in the Accountancy Age 35 Under 35.

Jon Wedge

While Jon’s client work focuses on the financial services sector, he also oversees the firm’s assurance service, as well as supporting the trainees following in his footsteps.

ELANA DIMMER

Elana joined us in 2017 as an ACA trainee, after graduating from Durham University where she had studied languages. She is now a manager in our assurance team.

ENQUIRY FORM

By submitting this form, the data provided will be used to perform your request according to our privacy policy.