In what appears to be the first published judgment in the Middle East concerning the misuse of artificial intelligence by a lawyer in litigation, the Qatar Financial Centre Civil and Commercial Court (the “Court“) has found a lawyer practising in Dubai in contempt of court for relying on fake case law generated by AI.
The judgment, delivered by the President of the Court, Lord Thomas of Cwmgiedd (former Chief Justice of England and Wales), serves as a timely reminder of the professional responsibilities lawyers bear when using AI tools. Lawyers remain responsible for verifying the authenticity of every authority they cite.
In this instance, the Dubai lawyer, acting for the Defendant, cited two legal authorities to support an application for an extension of time to serve the Defence. The Claimant (acting in person) subsequently wrote to the Defendant, requesting a copy of one of the authorities as he had not been able to find it. The Court Registry then ordered the Defendant to produce both authorities.
Initially, the Dubai lawyer claimed he was “unable to access judgment copies of the cases on the website of court.” The Court Registry then ordered him to file a witness statement explaining how he had come upon the two cases. He later admitted that “those references were made in error inadvertently due to copy-paste errors and wrong research … The judgments as cited were not available in the form implied, and the references were inadvertently included due to reliance on secondary sources / incomplete case law databases by mistake.” It ultimately emerged that the cases were hallucinations – fabricated by Google’s AI mode.
The Court’s decision
Lord Thomas acknowledged that this was the first occasion on which fake cases had been cited before the Court, although this issue had arisen globally – notably, in the United States, England and Wales and Canada. Given that all Court decisions are freely available online and the Court’s relatively small case database, he stressed that this error was inexcusable.
Article 35.2 of the Court Rules provides that a person commits a contravention where false or misleading information is provided to the Court. Contravention of Article 35 and contempt requires that the individual have the requisite intentional conduct. The Court was clear that the conduct here was not inadvertent, and there was plainly no reasonable excuse for it:
- any lawyer should be aware by the second half of 2025 of the necessity of checking that a case provided by an AI system is in fact a genuine decision of the Court. Failing to do so amounts to “reckless conduct, not caring whether there was or was not such a case”;
- the Dubai lawyer went one step further by telling the Court Registry that he was not able to access copies of the judgments on the Court’s website; and
- only after further inquiries did he admit the cases had been provided by Google.
Notwithstanding the seriousness of the conduct, the Court found that the apology tendered by the Dubai lawyer and the publication of this judgment was a sufficient penalty. Revealing his identity, Lord Thomas said, would inflict a “disproportionately harsh penalty given the nature and size of the legal profession practising within the GCC states and given that this is the first case where this has happened in this Court”.
Alongside the judgment, the Court released a Practice Direction for consultation. It is understood that the Practice Direction will make clear that if a lawyer fails to verify any citation, it will be considered a breach of conduct and sanctions imposed will include the full identification of the lawyer or law firm and reputational consequences.
Comment
AI has promised immense benefits for the legal industry – it can be an invaluable tool for lawyers, assisting with drafting bespoke clauses, summarising lengthy contracts and even conducting research when appropriately guided, in a fraction of the time. However, there lies the danger. In the short span of a year, lawyers have grown heavily reliant on AI, and many have stumbled by placing too much trust in its capabilities. The reality is that, at present, AI still “hallucinates” far too much to be a dependable tool for legal research. Its tendency to generate plausible but fictitious case law, due to the design of AI systems which seek to satisfy user prompts, remains a serious concern.
This judgment serves as a timely reminder of two things: first, the critical need to verify that any cases generated by AI are genuine; and second, the overarching responsibility lawyers bear for the accuracy of their work. In this instance, the QFC Court was more forgiving towards the Dubai lawyer, given that this was the first case of its kind before the Court, and the Dubai lawyer promptly apologised. However, far harsher repercussions for practitioners and law firms can be expected should this reoccur.
What is surprising is that such misuse continues to happen globally, despite growing awareness. There have already been more than half a dozen cases in the UK and approximately 50 in the US involving similar AI-related errors.
Mandatory AI training for lawyers has never been so critical. Separately, it will be interesting to see whether courts in other jurisdictions follow the QFC Court’s footsteps and release new rules or practice directions to mitigate these risks. Regulatory bodies will certainly also release guidance – the English Solicitors Regulation Authority has already published its “compliance tips for solicitors regarding the use of AI and technology”. It remains to be seen whether these steps will be sufficient to curb its misuse in this context, particularly as the technology continues to evolve at pace.
