AI and international arbitration – watching brief #1: the current landscape

Generative AI is transforming industries at an unprecedented pace, and international arbitration is no exception. At Freshfields, we believe that understanding and responsibly harnessing this technology is key to delivering the best results for our clients. In this new blog post series, we explain how generative AI is shaping the world of arbitration.

In this first post, Rohit Bhat, Joshua Kelly, Natalia Zibibbo, Guy MacInnes-Manby, Olivier André and Robert Stevens offer an introductory overview of the current landscape for the use of AI in arbitration, building on our trends for international arbitration in 2025 report (see here). Future posts will highlight the potential for specific reforms to procedural rules, the use (and misuse) of AI in arbitration, and emerging AI tools for arbitration, as well as other topics.

A fundamental question about AI’s application to arbitration is whether tasks that arbitrators already delegate – like legal research, administrative tasks, and partial drafting of an award – can also be delegated to AI, with the consent of the parties. 

As a consent-based and flexible form of dispute resolution, arbitration is well suited to become a key testing ground for AI:

• Arbitral tribunals are not necessarily bound by the procedural rules that may constrain the ability of domestic courts to incorporate technological developments into their processes. 
• Sophisticated businesses – as users of international arbitration – are likely to create competitive pressures that require international arbitration to use AI in a way that mirrors their own industries, to maximise efficiency. 
• Certain types of cases frequently submitted to arbitration (e.g., commodity disputes) use similar (and, sometimes, identical) language in their reasoning – e.g., when summarising legal principles applicable to a dispute – which may make it possible for AI to take a first pass at an award. 

Most arbitral institutions are still in the early stages of adopting procedural recommendations or guidelines for AI use, and the general trend is restrictive, rather than permissive, to ensure that arbitrators continue to independently analyse the law and the evidence relevant to a dispute. However, two key themes are emerging, concerning: (1) the importance of human oversight of any use of AI; and (2) disclosure of the use of AI by tribunals. For example:

• American Arbitration Association’s International Centre for Dispute Resolution (AAA-ICDR): The AAA-ICDR’s guidance prohibits an award being generated solely through AI-reasoning, and provides that AI tools should be used to “support—not replace—the arbitrator’s judgment and expertise”. It also recommends that arbitrators disclose the use of generative AI tools when such use “materially impacts” the arbitration process or the reasoning underlying their decisions. 
• Stockholm Chamber of Commerce Arbitration Institute (SCC): In its guidance note on AI, the SCC notes that in order to manage AI’s potential, AI should be qualified by “effective human oversight”, and that practitioners should “ensure any use of AI does not lead to any reduction in the quality of their decisions”. The SCC guidance note also encourages tribunals to disclose their use of AI in such tasks as research and analysis, to assist in ensuring that the tribunal “does not exceed its mandate”.
• Silicon Valley Arbitration and Mediation Center (SVAMC): The SVAMC’s guidance imposes on lawyers a “duty of competence or diligence” in their use of AI, while also encouraging disclosure on a case-by-case basis. (For more detail, see our earlier analysis of the SVAMC’s guidelines on AI in arbitration.)

Similar guidance has also been published by the China International Economic and Trade Arbitration Commission and the Chartered Institute of Arbitrators.

Several arbitral institutions are also taking proactive steps to promote the use of AI in arbitration. In particular:

• Leading institutions are encouraging discussion around AI amongst their users, and forming dedicated tasks forces and partnerships to explore and implement the use of AI in arbitration. For example, in 2025, the ICC Commission on Arbitration and ADR launched a task force to consider the use of AI. Similarly, the Hong Kong International Arbitration Centre (HKIAC) has partnered with Jus Mundi to offer free AI summaries of its procedural decisions using a tool that connects to the Jus Mundi database. The London Court of International Arbitration (LCIA) has also been facilitating discussions on AI, in the context of broader discussions around the current LCIA Rules.
• Institutions are also creating or adopting their own AI tools to improve efficiency. In October 2024, the AAA/ICDR began using artificial intelligence to assist case managers in selecting the most suitable arbitrator candidates from its roster during the nomination process. The tool, known as AAAi Panelist Search, is currently used internally, with plans underway to make it available to parties in the future. The AAA has also launched a beta version of an AI-powered drafting tool called ClauseBuilder, which can create bespoke arbitration clauses.
• Several institutions have taken steps to publicly signal their support for the responsible adoption of AI. The Singapore International Arbitration Centre noted that it "remains open to support the responsible and informed adoption of AI and technology in arbitration," while CIETAC has stated it would "vigorously promote" the application of AI in arbitration in its 2025 work plan.

As arbitral institutions continue with these initiatives, it is clear that AI's advancement in arbitration is unlikely to slow. Arbitrators will of course need to continue exercising independent judgment in their analysis of the law and the evidence, but this is unlikely to be a barrier to further adoption of AI.

For further information, see Freshfields’ recent report on how AI is expected to assist arbitration professionals with AI workflows.