At Dubai Arbitration Week 2026, institutional leaders from across the Middle East discussed the growing role of artificial intelligence in arbitration during a session titled ‘AI Center Stage’, organised by Jus Mundi in association with Al Tamimi & Company. The discussion focused on how arbitration institutions are adopting AI, the opportunities it presents for case administration and the challenges surrounding confidentiality, transparency and disclosure of AI use.

The discussion was moderated by Alexandre Vagenheim, Vice President of Institutional & Legal Affairs at Jus Mundi. The speakers included

● Hassan Arab, Regional Head of Dispute Resolution at Al Tamimi & Company;

● Khushboo Shahdadpuri, Partner at Al Tamimi & Company;

● Dalia Hussein, Director at CRCICA;

● Robert Stephen, Registrar at DIAC;

● Karam Farah, Counsel at the ICC;

● Jyothi Mani, Registrar at OCAC;

● Khaled Elgarhe, Chief Legal Counsel and Registrar at QICCA; and

● Arno Janssens, Head of Middle East at SIAC.

AI for case administration, document management

Opening the discussion, Alexandre Vagenheim, asked Robert Stephen what DIAC was already doing, or considering implementing, in its administration and institutional operations using AI. Robert Stephen explained that arbitration institutions are increasingly exploring AI for case management automation, document management, translation services and internal workflows. He said such tools could help institutions manage timelines, procedural updates, deadlines and large volumes of documents while providing arbitrators with a centralised and secure repository.

He also highlighted the possibility of using AI to summarise documents, identify key events and legal issues, and assist arbitrators in analysing case files. However, he emphasised that these were potential and developing applications, rather than functions already implemented across the board.

Dalia Hussein said CRCICA was taking a cautious approach to AI adoption, while continuing to work on case automation. She explained that the institution was using AI for internal research and reporting, including analysing its own case law and decisions on challenges to arbitrators, with the aim of providing users greater insight into institutional practices.

Arno Janssens similarly described SIAC’s approach as relatively conservative, particularly because of the confidentiality obligations attached to arbitration proceedings. He said the institution was examining which case-management tasks could be automated while ensuring that confidential information remained protected.

‘Clients care about cost, efficiency and accuracy’

Khushboo Shahdadpuri said law firms and their clients were increasingly using AI tools, but argued that arbitration users were primarily concerned with the quality and efficiency of institutional services rather than how much AI an institution used internally. She observed:

She pointed out that even minor administrative errors, such as incorrect calculations or dates, could affect users’ confidence in an institution. AI, she said, could help eliminate such basic errors and allow case managers to focus on more substantive aspects of proceedings.

Ms Shahdadpuri also referred to the development of an AI-powered arbitrator tool, which she said demonstrated how far AI applications in dispute resolution could potentially extend. She further questioned why arbitration institutions were not offering their own AI-powered tools, particularly where secure platforms could address confidentiality concerns.

Confidentiality remains a major concern

The speakers repeatedly returned to the issue of protecting confidential case information. Robert Stephen distinguished between AI tools integrated into a secure case-management system and the use of publicly available AI platforms.

He cautioned against uploading confidential case documents to open AI tools, stressing that institutions needed to ensure that case data was not exposed or accessed by unauthorised parties. He said secure, closed environments could offer useful applications for institutions, parties and arbitrators.

Khaled Elgarhe said QICCA’s use of AI remained limited to carefully controlled tasks, such as summarising data and correspondence. He emphasised that confidentiality was not merely a matter of institutional policy but could also involve legal obligations under national laws and institutional rules. He cautioned that improper disclosure of information to third-party AI tools could affect the validity of proceedings and arbitral awards.

Should arbitrators disclose their use of AI?

The panel also examined whether the use of AI by arbitrators should be disclosed to parties and whether institutional rules should expressly regulate such use.

Robert Stephen said DIAC was considering guidelines based on a traffic-light system to distinguish between AI uses that would be acceptable, those requiring disclosure and those that should not be permitted. He suggested that such guidance would need to evolve alongside technological developments.

Dalia Hussein favoured guidelines that would require parties to address AI use at the beginning of proceedings. She indicated that she would prefer disclosure obligations over incorporating detailed AI provisions directly into institutional rules.

Arno Janssens said SIAC had also chosen not to include express provisions on AI in its arbitration rules, noting that legal rules could lag behind technological developments. On disclosure, he suggested that the arbitration community was moving towards requiring disclosure where AI makes a material contribution to the arbitral process or the tribunal’s reasoning.

He emphasised that routine administrative tasks, such as scheduling and preparing procedural history, could be treated differently from the substantive reasoning and decision-making of an arbitral tribunal.

Karam Farah also highlighted the need to distinguish between administrative assistance and the delegation of judicial functions. He referred to the ICC’s ongoing work on an AI protocol, which could be discussed by parties and tribunals at the first case management conference. Such a protocol could address which uses require disclosure, what confidential information may be uploaded to AI tools, what outputs require independent verification and which uses should remain prohibited.

The panel broadly agreed that AI could improve arbitration administration, but its use must be accompanied by safeguards on confidentiality, transparency and human oversight.