Qatar: the QICDRC practice direction on artificial intelligence

Who it binds

It is issued under articles 38.2 and 26.2 of the Rules and Procedures of the QFC Civil and Commercial Court and the QFC Regulatory Tribunal respectively, and applies to “litigants, their legal representatives, and any legal practitioners who appear” before either body. The practice direction calls them Court Users, a defined term that reaches self-represented parties and foreign counsel alike.

Article 38.2 of the Rules and Procedures was read for this page. It is one sentence: “The President may issue Practice Directions or a Practice Guide in relation to the procedures to be followed before the Court, and the management and determination of Small Claims.” So the direction is not guidance offered to the profession. It is an exercise of the President’s rule-making power over procedure, and it is signed by the President himself.

If you are an international firm with a Qatar Financial Centre practice, or you advise on QFC-governed contracts, this binds your filings today.

It welcomes AI. Read that first.

“The use of Artificial Intelligence is to be welcomed in litigation in the QFC Court and RT with its potential to reduce cost and promote greater effectiveness.”, paragraph 4

That sentence carries a footnote to Jonathan David Sheppard v Jillion LLC [2025] QIC (E) 3 at paragraph 19. We have now read that judgment in full at the court’s own site, and the footnote is doing more work than it looks. Paragraph 19 is the passage in which the President, having just found a lawyer in contempt for citing fabricated authorities, says that AI is nonetheless to be welcomed and that a practice direction is required. The welcome and the contempt are in the same breath, in the same document. The case is set out below.

Paragraph 5 then closes the obvious loophole: nothing in the direction “alters, diminishes, or replaces a Court User’s existing ethical, professional, or legal duties, all of which continue to apply in full when using AI.”

The seven duties, by paragraph

• 7, 8Accuracy. Court Users remain responsible for the accuracy and reliability of what they submit. “It is a breach of duty to mislead the Court or submit materials containing fabricated or unverifiable legal citations.”
• 9, 10Know the failure modes. AI tools may generate false citations or misstate the law, and, uniquely on this record, “do not reliably distinguish between jurisdictions and may refer to irrelevant foreign legal principles.” No other court has named cross-jurisdictional contamination as a risk.
• 11Witness evidence. Court Users must not rely solely on AI output when drafting witness evidence. All AI-generated material must be verified before use.
• 12, 13, 14Confidentiality. “Confidential, privileged, or protected information must not be entered into AI tools, except where private, secure AI platforms are used.” Disabling chat history and using privacy-protective settings is encouraged. Breach may itself draw sanctions.
• 15, 16, 17Verification. All citations must be checked “against primary sources or recognised legal databases.” The Court will disregard submissions containing fictitious cases, legislation or factual assertions outside the evidence.
• 18Evidence. “AI tools must not be used to generate evidence. It is prohibited to ask AI to fabricate or tamper with evidence.” A flat prohibition, not a verification duty.
• 19, 20, 21eBundles. If AI is used to create or organise eBundles, the party must ensure correct indexing, pagination and relevance, conforming to Chapter 16 of the Maroon Book. Responsibility stays with the submitting party.

Deepfakes, and the disclosure rule

Paragraphs 22 and 23 do what the Federal Rules of Evidence in the United States have so far failed to do, in two sentences. Where visual or audio evidence is generated or enhanced using multimedia AI, the submitting Court User must clearly disclose the use and source of the technology. Any submission suspected of being a deepfake or manipulated without disclosure “may be excluded and investigated.”

Set that against the United States, where proposed Federal Rule of Evidence 707 was returned for further study in June 2026 and the deepfakes question with it. Qatar has had a rule since January.

On disclosure of AI use generally, the direction is deliberately calibrated:

• 24Pre-emptive declaration of AI use is not required unless the Court asks.
• 25The Court may decide that fairness or procedural good faith requires disclosure to other parties or to the Court.
• 26“Court Users must be prepared to identify portions of documents that used AI-generated content and explain how they verified the output. The Court may require an explanation or an affidavit.”

Paragraph 26 is the whole page in one sentence. You need not announce anything. You must be able to produce it on demand, on affidavit.

What misuse costs

Paragraph 27 lists four consequences, in escalating order: disregarding the affected pleadings or evidence in part or in whole; adverse costs orders; referral to professional regulators; and contempt of court proceedings where appropriate.

The first is the one to notice. Every other jurisdiction on this record punishes the lawyer. Qatar starts by discarding the client’s pleading, which is closer to Connecticut’s nonsuit than to an American fine.

The case behind the direction

Jonathan David Sheppard v Jillion LLC, neutral citation [2025] QIC (E) 3, case CTFIC0033/2025, judgment of 12 November 2025, Lord Thomas of Cwmgiedd sitting as President in the First Instance Circuit. Read in full at the court’s own site on 27 August 2026, and set out case by case on its own page in the sanctions record. Two lines of order:

“1. Contempt of court and breach of article 35.2 of the Rules and Procedures of the Civil and Commercial Court of the QFC proved. 2. There be no penal sanction.”

An employment claim. A lawyer practising outside Qatar, acting for the defendant, applied by email for an extension of time and cited two decisions of this Court: Al Khor International School v. Gulf Contracting Co. and Doha Bank v. KPMG. Neither exists. The claimant, acting in person, was the one who noticed: at 06:01 the next morning he emailed the Court asking for the neutral citation of Doha Bank, because he could not find it.

• First answerThe lawyer told the Registry he was “unable to access judgement copies of the cases on the website of court” and asked the Court to disregard them.
• Second answerOrdered to produce the cases or a witness statement with a statement of truth, he filed one attributing the citations to “copy paste errors and wrong research” and to “reliance on secondary sources / incomplete case law databases”. He apologised.
• Third answerAsked where he had found them, he produced screenshots of a Google search, explaining he had been helping a friend while in India, late at night.

The Court took the sequence itself as the evidence of intention. Paragraph 23, on the standard now expected of any lawyer:

“It is difficult to see how any lawyer could, by the second half of 2025 be under any doubt as to the necessity of checking that a case provided by an AI system is in fact a decision set out in the jurisprudence of the court. Proceeding therefore to refer to such a case without making such a check ordinarily amounts to reckless conduct, not caring whether there was or was not such a case.”

Then paragraph 24: the conduct “goes further”, because of what was said to the Registry when the citations were asked for. The Court did not accept that it was inadvertent. It was intentional conduct, with no reasonable excuse, and it was a contempt.

No penalty followed. The apology and the publication of the judgment were held sufficient, and the Court anonymised the lawyer, “not without very considerable hesitation”, on the ground that identification would be disproportionately harsh given the size of the profession practising across the GCC states, and given that this was the first case of its kind before this Court.

First, and last. The judgment closes by saying what the direction that followed would mean:

“… any citation of any case or other authority to this Court which has not been verified by an examination by the advocate of the case or other authority, will be considered a breach of the conduct required of all advocates before the Court and sanctioned accordingly. The sanctions will include the full identification of the lawyer or law firm and the consequent public disgrace.”

Read that against paragraph 26 of the practice direction. The duty is not to have used no AI. It is to have examined the authority yourself, and to be able to say so. The judgment names the standard; the direction names the evidence.

Two further details worth carrying. A draft of the practice direction was published for consultation at the same time as the judgment, so the instrument and the case are one act, not two. And the Court surveyed the same problem elsewhere before deciding: Mata v. Avianca is quoted at length, Ayinde at paragraphs 7 to 9, with Canadian and Australian matters noted alongside. The requisite intention test comes from its own earlier decision, Ferris v Sanguine Investment Management [2024] QIC (E) 1.

Sources

• Practice Direction on the Use of Artificial Intelligence (No. 1 of 2026), QICDRC PDF, read in full
• QICDRC, Legislation, Regulations and Procedures
• Court announcement, January 2026
• Jonathan David Sheppard v Jillion LLC [2025] QIC (E) 3, sealed judgment of 12 November 2025, read in full. Its case page, CTFIC0033/2025. An Arabic text is published alongside.
• Rules and Procedures before the Civil and Commercial Court of the QFC, for article 38.2, the rule-making power, and article 35, contravention and contempt.
• QICDRC User Guide, the Maroon Book, Chapter 16 on e-bundles, which paragraph 21 of the direction requires bundles to conform to.
• An accompanying Practice Guidance is referred to in the announcement. Searched again on 27 August 2026 across the court’s regulations page and its media centre: not found in this pass as a published document. The announcement describes it; we do not describe it from that description.

Related

• AI rules for lawyers, country by country
• Singapore, the other commercial-hub instrument
• Privilege and AI, which paragraph 12 speaks to directly
• Ontario Superior Court of Justice
• Tribunals Ontario, the other instrument that requires no pre-emptive declaration
• Sheppard v Jillion LLC, the judgment that produced this direction
• Mata v. Avianca and Ayinde v. Haringey, both relied on in Sheppard

Checked 5 August 2026 against the court’s own PDF, all 27 paragraphs, and re-checked 27 August 2026 against the Sheppard judgment, the Rules and Procedures and the User Guide. Information, not legal advice, and not advice on QFC practice.