During Hong Kong Arbitration Week 2026, A&O Shearman hosted a panel of senior lawyers, experts and arbitrators, moderated by Sheila Ahuja KC (A&O Shearman), in front of a packed audience to explore whether Generative AI is helping to make arbitrations faster and more cost-effective or, conversely, opening the door to guerilla tactics and unfair outcomes.
The panelists—Zoe Lau (Standard Chartered), Edward Taylor (A&O Shearman), Eric Ng (Prince’s Chambers), Jessica Pyman (Control Risks), Lydia Tang (BRG), and Fares Nowak (A&O Shearman)—shared their firsthand experiences and perspectives on how Generative AI is reshaping international arbitration in Asia. Several key themes emerged from the engaging panel discussion and audience Q&A.
1. Generative AI is already impacting key aspects of international arbitration
AI is already being extensively used in international arbitration. Existing core use cases include (1) document review and fact finding, (2) legal research, (3) preparing first drafts of correspondence, written submissions, expert reports, valuation models and procedural sections of arbitral awards as well as (4) identifying potential assets for enforcement.
2. Generative AI can deliver real efficiency gains. However, it is not risk-free nor cost-free and may not always simplify disputes
AI can now perform many basic, and increasingly complex, tasks in international arbitration quickly and efficiently. This has the potential to enhance access to justice and “level the playing field” among disputing parties.
However, its use (and potential misuse) is not without risks that could undermine arbitration’s legitimacy, including the “black box” problem, potential algorithmic bias as well as guerilla tactics such as prompt injection, fabricated evidence, and the rapid preparation of inaccurate and excessively lengthy written submissions.
Such risks are amplified in the context of documents-only expedited proceedings, where arbitral tribunals have less time and fewer opportunities to identify and remedy such issues. Indeed, the misuse of Generative AI during the arbitral process may lead to more arbitral awards being challenged before national courts.
Sophisticated Generative AI tools are also not cost-free (and such costs may increase overtime) and initial time savings may simply free up more time for counsel and experts to advance ever more complex arguments and case theories.
3. Human oversight, transparency and procedural safeguards can help to safeguard the arbitral process
Generative AI should not be used at the expense of human oversight and accountability, but its rapid evolution and ever-expanding use cases make effective regulation challenging. One important safeguard that arbitral tribunals are increasingly implementing is to require parties, counsel and experts to disclose how AI is being used.
Ultimately, the question for arbitration users and stakeholders is no longer whether AI will be used in arbitration (since that ship has sailed) but how it can be used responsibly without undermining the fairness, integrity, efficiency and finality of arbitral proceedings that arbitration users legitimately expect.
For further information, please feel free to contact Sheila Ahuja KC (Sheila.Ahuja@aoshearman.com) or Ed Taylor (Edward.Taylor@aoshearman.com).
