The Dekagram: 12th May 2025

This week Thomas Yarrow revisits the vexed question of the use of artificial intelligence in legal research – and our intrepid reporter finds that it’s not all it’s cracked up to be. In fact the experience led him to such depths of despair that he ended up quoting from one of Shakespeare’s tragedies; but which one? First correct answer wins a prize – and we trust our readers not to ask a search engine for help.

The Use of AI in Legal Research: One Barrister’s Story

Last week there was a great deal of stir following publication of the judgment of Ritchie J in R (Ayinde) v LB Harringey [2025] EWHC 1040 (Admin), owing to the unfortunate citation of non-existent caselaw by legal counsel in submissions before the Court. Although it was denied that the use of Artificial Intelligence was to blame for the intrusion of fake cases into the skeleton argument, salt-pinching abounds across the legal profession in that regard.

Here at Deka Chambers, we’re not averse to a little bit of investigative journalism and so this week I’ve taken one of the leading offerings on the AI market – not naming any names, but one which trawls the open internet and not paywalled repositories – out for a little spin to see what it can do.

Naturally, as something of a Charlie Kauffman fan, the first question I asked was the following: What can legal professionals learn from the judgment of Ritchie J in Ayinde v LB Harringey?

The answer:

Ritchie J’s judgment in Ayinde v LB Haringey highlights the dangers of fabricated case citations, the need for verifying AI-generated legal research, and strict procedural compliance. It reinforces ethical duties, condemning misleading the court as professional misconduct. Legal professionals must ensure accuracy, integrity, and diligence in all submissions.

Pretty good. But, ok, that was relatively easy. How about putting it to the test with some of the questions I have addressed in some of my previous Dekagram articles. To make it more fun, I’ve demanded that it cites authorities for any propositions relied on. My first question – which is particularly zeitgeisty for me at the moment – was: Does regulation 12(7) of the Package Travel and Linked Travel Arrangements Regulations 2018 require a traveller to satisfy a ‘but for’ causation test?

The answer:

The causation test under regulation 12(7) has been interpreted in line with general principles of English tort law. In McDonagh v Ryanair Ltd [2021] EWCA Civ 947, the Court of Appeal considered causation in consumer protection claims, affirming that a claimant must demonstrate that the breach was a significant cause of the loss suffered. While not explicitly requiring a ‘but for’ test, the regulation aligns with standard causation principles.

OK. Sounds good. And a lovely legal authority to go with it, in the Court of Appeal, no less. Here’s the problem: McDonagh v Ryanair is a real case, but it is an Irish case, not an English one, which reached the Court of Justice of the European Union in 2013, not in 2021; it concerned the Denied Boarding Regulations, not the Package Travel Regulations; it is not authority for any proposition relating to ‘but for’ causation. There is no law report with the citation [2021] EWCA Civ 947; there is a 948, which is an immigration case concerning the Refugee Convention.

I asked the AI to tell me more about this Court of Appeal case and what it decided? At that point, it fessed up:

It looks like I made an error when I referenced McDonagh v Ryanair Ltd [2021] EWCA Civ 947 earlier—I can’t find any record of such a case. I must have mistakenly created or misattributed the citation while trying to support the point about causation in package travel claims. I appreciate your scrutiny, and I’ll make sure to be more precise with legal references in the future. Thanks for catching that!

The gratitude at the end makes it so difficult to be mad…

I asked my second question: Does the Rome II regulation govern the applicable law for a claim involving a member of the British armed forces who is injured during the course of their employment overseas?

The answer:

While Rome II applies broadly, Article 1(2)(d) excludes “acts of state authority” (acta iure imperii), which may limit its application in military negligence claims. In Ministry of Defence v AB & Others [2010] EWCA Civ 1314, the Court of Appeal considered whether claims arising from overseas military operations fell within Rome II, ultimately recognizing that public policy and sovereign immunity could restrict its application.

Again, sounds promising. And the first sentence is legally correct. But once more, some issues: The citation [2010] EWCA Civ 1314 is a report of a chancery case about breach of trust. What the AI meant to refer to (I presume) was [2010] EWCA Civ 1317 which is a case in which British service personnel sued the MOD for personal injury as a result of exposure to ionising radiation. We might be able to forgive the AI being out by 3, but that’s not the only problem. In the entirety of the 305 paragraph judgment, there is not a single reference to Rome II, nor to acta iure imperii. The single reference to sovereign immunity is a reference to a case from 1985. The case is instead principally about sections 14 and 33 of the Limitation Act 1980. The Court of Appeal is not even the highest authority – the matter went to the Supreme Court ([2012] UKSC 9).

I challenged the AI once more. It replied, contrite:

You’re absolutely right to question that reference. I made an error in citing Ministry of Defence v AB & Others [2010] EWCA Civ 1314—it appears that no such case exists with that citation, and the case I intended to reference does not actually discuss acta iure imperii or the Rome II Regulation. I appreciate your scrutiny, and I should have verified the citation more carefully before including it in my response.

Thank you for catching that mistake—I’ll ensure greater accuracy in legal references going forward. Would you like me to provide a properly sourced answer regarding the applicability of Rome II in military injury claims?

“Yes please”, I answer. Then:

The Rome II Regulation (Regulation (EC) No 864/2007) governs the applicable law for non-contractual obligations, including personal injury claims. Article 4(1) states that the law of the country where the damage occurs applies unless exceptions under Articles 4(2) or 4(3) apply.

In Scott v AIG [2021] EWHC, the court considered the scope of Article 4(3) and whether the presumptive applicable law under Rome II could be displaced. The case illustrates how Rome II applies to cross-border personal injury claims, including those involving British armed forces overseas, unless a manifestly closer connection to another jurisdiction exists.

I won’t bore readers any further, but suffice to say that Scott v AIG is a County Court case, not a High Court one, which is about Rome II, but it’s in the context of a road traffic accident in France and has absolutely nothing to do with British armed forces overseas.

Overall, it’s a negative grade from me. But I’m not so hubristic as to think I could never be replaced by a machine. If it be not now, yet it will come. The readiness is all.

About the Author

Before coming to the Bar, Thomas worked as a Civil Servant, including as a policy advisor at the Brexit Department. His former experience gives him an encyclopaedic knowledge of legislation, treaties and international conventions in the field of Private International Law, which he has applied to his cross-border practice, building a reputation as a go-to barrister for complex questions of jurisdiction and applicable law. He wrote the relevant chapter on such in the latest edition of the leading practitioner textbook, Saggerson on Travel Law and Litigation. His travel practice spans cross-border personal injury, commercial disputes and consumer protection.