What was actually announced
The news is not a product launch but a professional position. According to a press statement received by Al-Bawaba Al-Taqaniya, the Sharjah International Commercial Arbitration Centre («Tahkeem») held a session within Dubai Arbitration Week 2026 titled “The Arbitrator of the Future: Human, AI, or Hybrid?”, at the Waldorf Astoria in the Dubai International Financial Centre. The panel examined whether AI can assist in analysing documents and legal evidence, managing cases, and what that means for international arbitration practice. It also addressed transparency, accountability, and preserving the arbitrator’s independence and discretionary authority in issuing awards and decisions.
The speakers were Dr Ahmed Saleh Al-Ajla, director of «Tahkeem»; Dr Zeina Obeid, partner at Obeid & Partners; and Dr Jalal El-Ahdab, partner at FIDAL in Paris. The discussion was moderated by Nassif Bou Malhab, co-head of the international arbitration group at Greenberg Traurig, and Mohieddine Al-Banna, founder and CEO of QAF Legal.
Al-Ajla’s framing is the substance worth reporting: technology forces a rethink of traditional working mechanisms, and AI can be used to analyse information, accelerate procedures and improve case-management efficiency — but the core of the arbitral process remains legal expertise and the arbitrator’s independent professional judgement, including understanding the legal context, facts and circumstances of each dispute. The challenge, he said, is deploying these technologies responsibly and in an integrated way while preserving guarantees of fairness, impartiality and confidentiality, and developing regulatory frameworks and professional practices so that the award remains grounded in independent human assessment and professional responsibility.
How this differs from the previous debate
Two years ago, the arbitration conversation about AI was largely defensive: disclosure obligations, whether generative tools had been used, and how to sanction parties who relied on fabricated authorities. This session moves the frame from suspicion to architecture. The question is no longer “may we use AI?” but “where in the process, under what safeguards, and who remains accountable?” That is a meaningful shift, and it comes from an institution rather than a vendor.
It also differs from how legal-tech companies typically present the same capabilities. Commercial providers sell speed in review and drafting; «Tahkeem»’s panel put confidentiality, data protection and arbitrator independence at the centre of the discussion, alongside transparency and accountability. In other words, the institutional agenda is setting the conditions of adoption rather than reacting to tools already in use.
The source gives no comparative figures, benchmarks or product specifications, so no table of numbers would be honest here. The distinction is conceptual: the debate has moved from whether machines can read a case file to whether a tribunal can delegate part of that reading without eroding the human judgement on which enforcement depends.
Three concrete uses that follow from the discussion
1. Document and evidence analysis in large commercial disputes. The panel explicitly named analysing documents and legal evidence as a candidate use. A realistic example: a construction dispute with 40,000 pages of correspondence. A tribunal’s assistant, or a party’s counsel, uses a review tool to cluster documents by issue and surface the handful of letters where notice periods were disputed, then verifies each against the original. The tool narrows the reading; the arbitrator still decides what the letters mean. If you are building that workflow, the discipline described in AI Hallucinated Citations: Verify Legal Research Before Court applies with equal force in arbitration.
2. Case management and procedural administration. Al-Ajla referred to accelerating procedures and improving the efficiency of dispute management. A realistic example: a case-management system that tracks procedural deadlines, drafts standard procedural orders for the tribunal’s review, and flags conflicts or missing filings before a hearing. The gain is administrative, not adjudicative — which is precisely why it is the least contested use.
3. Hybrid support for the sole arbitrator or small firm. The session’s title suggests the answer to “human, AI or hybrid?” is the third option. A realistic example: a sole arbitrator in a smaller Gulf-seated case uses an assistant tool to summarise each party’s submissions into a common chronology, then rewrites the analysis personally and records in the procedural history that assistance was used. For practitioners building such workflows deliberately, see AI agents for lawyers.
Who and what this affects
The immediate audience is arbitrators, arbitral institutions and counsel practising in the Gulf and in international commercial arbitration. But the practical effects reach further.
Institutions will need procedural rules and guidance on disclosure of AI assistance, retention of prompts and outputs, and the boundaries of what may be delegated to a tool. «Tahkeem»’s stated emphasis on developing regulatory frameworks and professional practices signals that guidance is being contemplated, not merely discussed.
Parties and counsel face a confidentiality question that is not theoretical: uploading a confidential submission or an evidentiary bundle into a third-party tool creates a data-protection and privilege exposure that no efficiency gain offsets if it is mishandled. The panel’s attention to confidentiality and data protection reflects that risk.
Founders and legal-tech builders should read the session as a specification list: audit trails, data residency, human-in-the-loop design, and outputs that can be verified rather than trusted. Arbitrator independence is a product requirement, not a marketing line.
The award itself is the ultimate stake. Al-Ajla’s insistence that the decision remain based on independent human assessment and professional responsibility is a statement about enforceability: an award that reads as machine-generated invites challenge.
What to watch: confidentiality, accuracy and the limits of use
Four issues deserve attention as this debate matures.
Confidentiality and data protection. Arbitration is confidential by design. Any tool that processes party material outside controlled environments must be assessed against that obligation before, not after, adoption.
Accuracy and accountability. The panel discussed transparency and accountability explicitly. In practice that means knowing which model produced which output, on what material, and who reviewed it. Without that record, accountability dissolves.
Independence and discretion. The arbitrator’s discretionary authority is not a sentimental attachment; it is the reason parties chose arbitration. Delegating evaluation, rather than retrieval or organisation, is where the line sits.
Intellectual property and provenance. Training data, ownership of generated summaries and the copyright status of machine-assisted drafts remain unsettled, and they matter when a procedural document is later relied upon. The source does not address IP directly, so treat this as an open issue the profession has not yet resolved rather than a settled rule.
What the source notably does not provide: any product names, prices, or quantitative benchmarks. Anyone claiming otherwise is adding to the story.
How and when to engage with this
The source describes a completed session at Dubai Arbitration Week 2026, held at the Waldorf Astoria in the Dubai International Financial Centre, with no announced follow-up instrument, rule change or publication. Practitioners cannot “try” a debate.
What they can do now, on the basis of what was said: review internal workflows to identify where AI is already being used informally; document that use; restrict it to analysis, organisation and administration rather than evaluation; and press institutions for guidance on disclosure. If you advise clients on positioning around these questions, the disclosure discipline overlaps with what is discussed in legal marketing only in one respect — consistency between what you say publicly and what you actually do.
Watch «Tahkeem»’s publications and any procedural guidance emerging from the centre for the practical form this discussion takes. Until then, the session’s value is as a statement of institutional intent: adoption is welcome, delegation of judgement is not.
Sources
- البوابة العربية للأخبار التقنية: «تحكيم» يناقش دور الذكاء الاصطناعي وحدوده في تسوية المنازعات
Frequently asked questions
What exactly did «Tahkeem» announce?
It did not announce a product or rule change. According to a press statement received by Al-Bawaba Al-Taqaniya, the Sharjah International Commercial Arbitration Centre held a session at Dubai Arbitration Week 2026 titled “The Arbitrator of the Future: Human, AI, or Hybrid?”, discussing AI uses in document and evidence analysis and case management, and the challenges of confidentiality, data protection and arbitrator independence.
Did the panel conclude that AI can replace arbitrators?
No. Dr Ahmed Saleh Al-Ajla, director of «Tahkeem», said the core of the arbitral process remains legal expertise and the arbitrator’s independent professional judgement, and stressed that the award should remain grounded in independent human assessment and professional responsibility. The framing of the session — human, AI or hybrid — points toward assisted rather than autonomous decision-making.
What are the main risks raised in the discussion?
The panel focused on confidentiality and data protection when handling case material, transparency and accountability for how AI is used, and preserving the arbitrator’s independence and discretionary authority. Al-Ajla also emphasised developing regulatory frameworks and professional practices so that efficiency gains do not undermine guarantees of fairness, impartiality and confidentiality.
Want to apply this to your own work or your team? Tell me where you stand, and I will reply personally.
Get in touch
