← All case files
verified deployment legal · AE · cross

ADGM Court (Heath J.): AED 282,508 indemnity costs against MIO Legal Consultants for AI-fabricated authorities in a defence pleading

In a costs judgment dated 18 December 2025, the Abu Dhabi Global Market Court of First Instance ordered MIO Legal Consultants LLP to pay AED 282,508 on the indemnity basis after its defence relied on unverified AI legal research citing cases that did not exist. The court found the firm's failure to verify the AI research reckless and a breach of the ADGM Courts Rules of Conduct 2016.

MetricBeforeAfter
MIO Legal Consultants LLP ordered to pay AED 282,508 costs on the indemnity basis (Wasted Costs Application)
Court found the firm's failure to verify AI legal research reckless and a breach of the Rules of Conduct 2016
Defence cited authorities that did not exist, incl. 'Allied Dunbar [1985]' and 'Johnston v Moreton [2014] EWHC 1219 (Ch)'

The problem

A commercial dispute between Arabyads Holding Limited (claimant) and its former employee Gulrez Alam Marghoob Alam (defendant) over an employee share-option arrangement came before the Abu Dhabi Global Market (ADGM) Court of First Instance, Commercial and Civil Division. The defendant’s lawyers, MIO Legal Consultants LLP, filed a Defence that the court described as extraordinarily lengthy: “The narrative alone ran to some 327 paragraphs (47 pages) compared with the 58 paragraphs contained in the Claim; with exhibits added the total length was 233 pages” source. Opposing counsel found that a number of the authorities cited in the Defence did not exist, carried incorrect citations, or did not stand for the propositions advanced source.

What was built

The Defence was prepared using generative-AI legal research. The court found that “AI (in one form or another) was used by MIO to undertake legal research for the purposes of preparing the Defence,” and that the defects “have all the hallmarks of the hallucinatory results that can occur when AI is used as a research tool” source. The specific AI product is not named in the judgment. Among the fabricated citations, the court identified two cases that “do not exist”: “Two cases were cited (at paragraphs 236 and 313(b) of the Defence respectively) which do not exist. These were ‘Allied Dunbar [1985]’ and ‘Johnston v Moreton [2014] EWHC 1219 (Ch)’” source. Indian legal-news outlet 24law reported the same episode as a firm sanctioned “after finding that the firm filed pleadings containing fabricated, mis-cited and misapplied legal authorities” source.

The outcome

The AI-linked sanction: AED 282,508 against the firm, on the indemnity basis. On the Wasted Costs Application, Justice Paul Heath KC ordered: “In relation to the Claimant’s Wasted Costs Application, MIO Legal Consultants LLP shall pay the Claimant its costs in the sum of AED 282,508 which have been summarily assessed on the indemnity basis” source. That figure and the indemnity basis are independently confirmed by two global-firm case analyses: Morgan Lewis reports the firm “were ordered to pay AED 282,508 to the Claimant in respect of the costs” source, and Clyde & Co that “the Firm was ordered to pay costs on an indemnity basis in the sum of AED 282,508 to the Claimant” source. The court held that “MIO’s failure to verify whatever legal research was undertaken through AI, was, I find, reckless, and amounted to a breach of the Rules of Conduct” source — a finding both firms echo: Morgan Lewis records that the conduct “amounted to a breach of ADGM Court Rules of Conduct 2016” source, and Clyde & Co that it “amounted to a breach of the Conduct Rules” source. The order was directed at the firm, not the individual lawyers. (A separate AED 245,000, on the standard basis, was ordered against the defendant personally on unrelated costs applications — not part of the AI-linked sanction source.)

A note on sources

The two criticals — the AED 282,508 indemnity award and the reckless-breach finding — each rest on the Tier-1 sealed judgment plus two mutually independent Tier-2 firm analyses (Morgan Lewis and Clyde & Co), so they are not single-sourced. The weakest load-bearing source here is [4], 24law.in, a regional Indian legal-news aggregator whose value-add is the ₹63-lakh currency gloss; it corroborates the sanction and the fabrication finding but is not relied on for any figure the primary judgment does not already state on its face.

How this was verified

  • Method: The Tier-1 primary is the ADGM CFI costs judgment (Justice Paul Heath KC, 18 December 2025, neutral citation [2025] ADGMCFI 0032, case ADGMCFI-2025-165), downloaded live and byte-identical from the official assets.adgm.com store and stored in sources/. Every quoted figure and finding was re-grepped verbatim against that judgment text this session; the two independent firm analyses and the 24law report were re-checked against local captures.
  • Conflation guards: the AED 245,000 standard-basis order is against the defendant personally, not the AI sanction; AED 785,581 was the total claimed; the AI product is unnamed in the judgment (the story asserts no ChatGPT/Gemini); the sanction falls on the firm (MIO), not the individual lawyers.
  • Date verified: 2026-08-18.
  • Status: PENDING — publish-ready; a court-origin adjudicated record with no private client to confirm. Open only: a human confirming on the ADGM Courts docket that the 18 December 2025 costs judgment is final/unappealed and whether the AED 282,508 was paid or enforced. Not verified — the green badge is the checker’s to grant against the public record.

Sources

  1. ADGM Court of First Instance · Arabyads Holding Ltd v. Gulrez Alam Marghoob Alam — Costs Judgment (Justice Paul Heath KC), [2025] ADGMCFI 0032, case ADGMCFI-2025-165 · 18 December 2025 · Tier 1 (court-origin primary) · https://assets.adgm.com/download/assets/ADGMCFI-2025-165+-+Costs+Judgment+-+Justice+Paul+Heath+KC+-+18122025+SEALED.pdf/68b3146edcba11f0a1f22292341fda95
  2. Morgan Lewis · ADGM Court’s Judgment Serves as a Warning Regarding Proper Use of AI · February 2026 · Tier 2 (independent firm analysis) · https://www.morganlewis.com/blogs/shiftingsandsoflaborlaw/2026/02/adgm-courts-judgment-serves-as-a-warning-regarding-proper-use-of-ai
  3. Clyde & Co · The use of AI in employment litigation · January 2026 · Tier 2 (independent firm analysis) · https://www.clydeco.com/en/insights/2026/01/the-use-of-ai-in-employment-litigation
  4. 24law · Abu Dhabi Court Slaps ₹63 Lakh Costs On Law Firm Over Unverified AI-Generated Citations · 2026 · Tier 2 (regional legal-news press) · https://24law.in/story/abu-dhabi-court-slaps-63-lakh-costs-on-law-firm-over-unverified-ai-generated-citations

Generative AI legal-research tool (unnamed in the judgment)

Verification record
Status
verified
Method
Tier-1 primary is the ADGM CFI costs judgment (Justice Paul Heath KC, 18 Dec 2025, neutral citation [2025] ADGMCFI 0032), served live and byte-identical from the official assets.adgm.com store; corroborated by independent commentary (Morgan Lewis, Clyde & Co) and legal press (24law.in).
Verified on
2026-08-24
Provider
Abu Dhabi Global Market (ADGM) Court of First Instance — Justice Paul Heath KC
Client
Arabyads Holding Limited v. Gulrez Alam Marghoob Alam (MIO Legal Consultants LLP sanctioned) · Legal / courts (adjudicated AI-hallucination sanction)
Disclosure
named
Questions this file answers
Why did the ADGM Court sanction MIO Legal Consultants?

In a costs judgment dated 18 December 2025, the Abu Dhabi Global Market Court of First Instance ordered MIO Legal Consultants LLP to pay AED 282,508 on the indemnity basis after its defence relied on unverified AI legal research that cited cases which did not exist. Justice Paul Heath KC found the failure to verify reckless and a breach of the ADGM Court Rules of Conduct.

Which fabricated cases were cited in Arabyads v Alam?

The court identified two authorities, at paragraphs 236 and 313(b) of the defence, that do not exist: 'Allied Dunbar [1985]' and 'Johnston v Moreton [2014] EWHC 1219 (Ch)'. The judgment found the defects had all the hallmarks of hallucinatory results from using AI as a research tool; the specific AI product was not named.