TL;DR — Qatar's QICDRC Practice Direction No. 1 of 2026 does not require lawyers to declare that they used AI. Paragraph 24 says the opposite: pre-emptive declaration is not required unless the court asks for it. What it does require, without condition, is that every citation be checked against primary sources or a recognised legal database. It binds the QFC Court and Regulatory Tribunal only — not Qatar's onshore courts — and it was not the region's first instrument: the DIFC Courts published guidance two years earlier, and theirs is stricter on disclosure.
The short answer
On 6 January 2026 the Qatar International Court and Dispute Resolution Centre issued its Practice Direction on the Use of Artificial Intelligence (No. 1 of 2026). It runs to 27 paragraphs over three pages and is signed by two people: Lord Thomas of Cwmgiedd, President of the Qatar International Court, and Sir William Blair, Chairman of the Qatar Financial Centre Regulatory Tribunal.
It has been summarised widely, and much of the summarising is wrong. The most common version — that Qatar now forces lawyers to flag AI-generated content and swear to it by affidavit — is not what the document says. The direction is more precise and, in one respect, more permissive than the guidance already operating one country over.
Key facts
- Issued 6 January 2026. The instrument contains no commencement clause, so there is no separate entry-into-force date to cite.
- It binds the QFC Civil and Commercial Court and the QFC Regulatory Tribunal. It does not reach Qatar's onshore national courts.
- It binds 'Court Users' — litigants, their legal representatives, and practitioners appearing. That includes litigants in person.
- Verification is unconditional. Disclosure of AI use is not, except for multimedia evidence.
- It came out of a real case: Sheppard v Jillion LLC [2025] QIC (E) 3, decided 12 November 2025.
What it actually requires, paragraph by paragraph
The direction is short enough to read in full, and worth doing so. This is the operative content, mapped to its own numbering.
| Paragraph | What it requires |
|---|---|
| 5 | Existing ethical, professional and legal duties continue to apply in full |
| 7 | Court Users remain responsible for the accuracy and reliability of what they submit |
| 8 | Misleading the court, or submitting fabricated or unverifiable citations, is a breach of duty |
| 11 | Witness evidence must not rely solely on AI output; AI material must be verified before use |
| 12 | Confidential, privileged or protected information must not be entered into AI tools, except on private secure platforms |
| 14 | Misuse of AI in breach of confidentiality may result in sanctions |
| 15 | Court Users must review and verify the factual and legal accuracy of AI-generated material |
| 16 | All citations must be checked against primary sources or recognised legal databases |
| 17 | The court will disregard submissions containing fictitious cases, legislation or unevidenced assertions |
| 18 | AI must not be used to generate evidence; fabricating or tampering with evidence is prohibited |
| 20 | Responsibility for the accuracy of eBundles stays with the submitting party |
| 22 | AI-generated or enhanced visual and audio evidence must be disclosed, with its source |
| 23 | Suspected deepfakes may be excluded and investigated |
| 24 | Pre-emptive declaration of AI use is not required unless the court specifically asks |
| 26 | Court Users must be prepared to identify AI-assisted passages and explain how they verified them; the court may require an explanation or an affidavit |
| 27 | Sanctions: disregarding pleadings or evidence, adverse costs, referral to professional regulators, contempt proceedings |
Three things most summaries get wrong
1. It does not require you to declare that you used AI
Paragraph 24 is explicit: pre-emptive declaration of the use of AI is not required unless the court specifically asks for it. Paragraph 25 preserves the court's power to decide that fairness or procedural good faith requires disclosure in a given case. So disclosure is a power the court holds, not a box you tick when filing.
The single exception is multimedia. Paragraph 22 makes disclosure unconditional where visual or audio evidence has been generated or enhanced using AI, and paragraph 23 lets the court exclude and investigate anything it suspects is a deepfake.
2. The affidavit is contingent, not standing
Paragraph 26 requires Court Users to be prepared to identify which portions of a document used AI-generated content and to explain how they verified the output. The court may then require an explanation or an affidavit. 'May' is the operative word. There is no affidavit to file alongside a pleading, and no certification form of the kind some United States federal judges have adopted in their standing orders.
3. It was not the region's first instrument
The Dubai International Financial Centre Courts issued Practical Guidance Note No. 2 of 2023 on large language models and generative AI in proceedings, in force from 21 December 2023 — two years ahead of Qatar. It is self-described as guidance rather than a binding rule, but it uses mandatory language and hangs on binding provisions: the DIFC Rules of Court on statements of truth and contempt, and the Mandatory Code of Conduct for Legal Practitioners.
And on disclosure specifically, the comparison runs opposite to the received wisdom. The DIFC asks parties to declare at the earliest possible opportunity if they have used or intend to use AI-generated content. Qatar asks for nothing unless the court raises it. On this one point, Qatar is the more permissive of the two.
Who it binds
The direction covers proceedings before the QFC Civil and Commercial Court — both its first instance and appellate circuits — and the QFC Regulatory Tribunal. It does not extend to Qatar's onshore national courts, which is the error in the most widely circulated law-firm summary of it. A note published in February 2026 stated that the direction applies to all litigants and representatives appearing before courts or tribunals 'in the country'. It does not.
It is also not limited to QFC-registered entities. The QFC Court's jurisdiction reaches disputes between QFC entities and parties established elsewhere in Qatar, among other heads, so a party with no QFC registration can find itself before the court and bound by the direction as a Court User.
Try HAQQ AI Free
Experience AI-powered legal drafting and research
The case behind it
In Sheppard v Jillion LLC [2025] QIC (E) 3, decided on 12 November 2025, the court found contempt and a breach of article 35.2 of its own rules, and imposed no penal sanction. The lawyer involved was not named; the court anonymised him. A draft of the practice direction was published for consultation alongside the judgment, and the final text followed seven weeks later.
The judgment had signalled a harder line than the direction ultimately took. At paragraph 26 it forecast that sanctions would include full identification of the lawyer or firm and the consequent public disgrace. No naming provision appears in the direction as issued.
Qatar, the DIFC and the ADGM compared
| Forum | Instrument | Disclosure duty | Nature |
|---|---|---|---|
| Qatar — QFC Court and Regulatory Tribunal | Practice Direction No. 1 of 2026 (6 Jan 2026) | Only if the court asks; unconditional for multimedia | Practice direction |
| Dubai — DIFC Courts | Practical Guidance Note No. 2 of 2023 (21 Dec 2023) | Declare at the earliest possible opportunity | Self-styled guidance, tied to binding rules |
| Abu Dhabi — ADGM | No AI-specific instrument | Not specified | Existing conduct rules, enforced |
| Saudi Arabia | None located | Not specified | Rules of Professional Conduct contain no AI provision |
The ADGM row is the one worth pausing on. It has no AI practice direction at all, and it has still produced the region's most expensive AI citation failure. In Arabyads Holding Limited v Alam [2025] ADGMCFI 0032, decided 18 December 2025, the Court of First Instance ordered a firm to pay AED 282,508 in wasted costs on the indemnity basis for filing unverified AI-generated authorities — under conduct rules written in 2016, years before the tools existed.
The pattern across all four: nobody needed a new rule to sanction an unverified citation. The duty to check what you file was already there. What the new instruments add is specificity about how you will be asked to prove you checked.
What this means for a Qatar-facing practice
- Build the verification record as you work, not afterwards. Paragraph 26 asks you to be prepared to explain how you verified — which is a question about process, and answerable only if the process left a trace.
- Check every citation against a primary source or a recognised legal database. Paragraph 16 admits no exception and no materiality threshold.
- Keep client confidential material out of general-purpose AI tools. Paragraph 12 permits private, secure platforms and nothing else.
- Disclose AI-generated or enhanced audio and visual evidence without waiting to be asked. This is the one unconditional disclosure duty in the document.
- Do not treat the absence of a declaration requirement as the absence of a duty. Verification is mandatory; only the declaration is discretionary.
Key takeaways
- Qatar's direction requires verification of every citation, but does not require declaring AI use unless the court asks.
- The affidavit in paragraph 26 is a power the court may exercise, not a filing requirement.
- Scope is the QFC Court and Regulatory Tribunal, not Qatar's onshore courts.
- The DIFC Courts got there two years earlier and ask for earlier disclosure.
- The ADGM has no AI rule and has still imposed AED 282,508 in costs for unverified AI citations.
