Part 5 - AI for legal professionals: AI in arbitration
Arbitration is characterised by party autonomy and procedural flexibility, often involving a cross-border dimension. While many AI applications, such as legal research, document review and predictive analytics, are not specific to arbitration, arbitration presents unique opportunities for leveraging AI in ways that reflect its distinctive procedural flexibility and international scope. The fifth instalment in our series on AI for legal professionals[1], this article focuses on the ways in which AI is being, and may increasingly be, deployed in arbitration.
The increasing role of AI in arbitration
At first glance, the uses of AI in arbitration mirror those in litigation. Tools can be deployed by parties to arbitration for document review, e-disclosure, legal research and even (with oversight) the drafting of certain documents. Properly deployed, AI can enhance consistency, reduce the risk of oversight and facilitate greater access to arbitration by reducing the costs associated with manual document review and translation. Further, predictive analytics[2] could be applied not just to forecast case outcomes but to assess the likelihood of enforcement challenges in particular jurisdictions. However, arbitration is a distinct process and certain features of arbitral proceedings give rise to more specific use cases for AI.
Tribunals and arbitral institutions are also increasingly turning to AI to assist with tasks, such as scheduling, drafting procedural orders and the collation of issues lists. These functions may be largely administrative but they can materially improve efficiency in complex arbitrations, especially where resources are stretched or parties are located across multiple jurisdictions and time zones.
Arbitrator selection
Unlike judicial assignment in litigation, arbitrator selection is party-driven. While parties have traditionally relied on personal networks, institutional lists or manual review of arbitrator profiles to select arbitrators, sophisticated matching tools present parties with one of the most notable potential uses of AI in arbitration. AI algorithms can be used to analyse vast datasets, including prior awards (to the extent they are publicly available), subject matter expertise, availability, language skills and even patterns in decision-making, to recommend suitable candidates for appointment.
Some arbitral institutions have begun piloting platforms where parties input case parameters and receive ranked lists of potential arbitrators generated by AI models. For example, arbitral institutions such as the International Chamber of Commerce (ICC) and Singapore International Arbitration Centre have begun piloting platforms that leverage machine learning to match arbitrators with cases based on nuanced criteria provided by parties. These systems can also cross-reference disclosed relationships against public databases to flag potential conflicts of interest, a process that is increasingly automated and far more comprehensive than traditional manual checks.
Translation tools
The international nature of arbitration means that parties often speak different languages and accurate translation is essential. AI-powered translation services can automatically translate documents held on e-discovery platforms which can assist with the review of documents in different languages in complex cross-border arbitrations.
A word of caution, however: as with legal research and drafting, when it comes to the translation of legal documents, human translators with legal and jurisdiction-specific experience remain superior to machine translations which still require human oversight and the tribunal may well require translations certified by a professional for use in submissions.
Risks and challenges
While AI has clear potential to enhance the arbitral process, we ought to be aware of the risks of using AI. Confidentiality, for example, is paramount in arbitration. Uploading sensitive documents to third-party AI platforms may risk breaching confidentiality obligations or undermining enforceability. Practitioners must carefully scrutinise the terms of use of AI tools and consider data protection regimes in multiple jurisdictions.
Another challenge is legitimacy. If AI tools are used without transparency or if parties feel disadvantaged by another’s reliance on opaque algorithms, there is a risk that awards could be challenged on procedural grounds. Equally, arbitrators must not outsource their decision-making to machines: while AI may assist in processing material, the ultimate responsibility for reasoning and judgment remains with the tribunal.
Finally, while challenges such as transparency, accountability, bias and data privacy are common to the use of AI in arbitration and litigation alike, concerns about opacity may be more pronounced when decisions are influenced by proprietary algorithms not subject to public scrutiny. Institutions are responding with guidelines requiring the disclosure of any algorithmic involvement in decision-making processes and some have established audit trails for AI-generated recommendations related to arbitrator selection or the drafting of awards.
CIArb guidelines and other guidance
The Chartered Institute of Arbitrators’ (CIArb) recent Guideline on the Use of AI in Arbitration (2025) provides a comprehensive, though non-binding, framework for managing the opportunities and risks of AI in arbitration. The guidelines encourage proactive agreement between parties on the permissible uses of AI, and emphasise that tribunals may regulate parties’ AI deployment with a view to preserving the integrity of arbitral proceedings and ensuring the validity and enforceability of any ensuing awards.
Arbitrators are urged to consult parties before using AI tools themselves, disclose any intended use, and never delegate decision-making to AI[3]. The guidelines emphasise that tribunals remain ultimately responsible for the reasoning and conclusions of awards, regardless of technological assistance. They also envisage tribunals appointing independent AI experts to evaluate contested tools where appropriate.
A notable concept introduced in the guidelines is “High Risk AI Use”, which requires enhanced scrutiny where AI is deployed in ways that could affect confidentiality or procedural integrity of the decision-making process. In this respect, the guidelines echo the EU AI Act, which categorises AI systems intended for use by judicial authorities and alternative dispute resolution bodies as “high risk”.
In addition to the CIArb guidelines, other leading arbitral institutions are beginning to articulate their own approaches to AI, demonstrating that this is a cross-institutional priority. For instance, the ICC’s Commission on Arbitration and ADR has established a dedicated Task Force on Artificial Intelligence in Dispute Resolution. This task force is tasked with producing practical guidance for parties and arbitrators, while preserving core arbitration values such as party autonomy, fairness, and confidentiality.
For practitioners, guidance on AI is both practical and symbolic. The CIArb guidelines in particular offer a ready-made procedural language, template agreements and procedural orders that can help parties and tribunals proactively manage AI use – and they signal that lawyers and arbitrators are expected to engage with AI in an informed, transparent and cautious manner. They also remind practitioners that institutions and regulators will increasingly scrutinise AI’s role in proceedings, meaning early consideration of AI protocols is now a best practice rather than an optional extra.
Comment
As arbitral institutions continue experimenting with new technologies, practitioners should remain vigilant about both opportunities for increased efficiency and challenges related to confidentiality and enforceability. With guidelines now emerging, arbitration stands at a pivotal moment: to embrace technological innovation while safeguarding the trust that underpins its role in international dispute resolution.
This article is the last instalment in our series on AI for legal professionals for 2025. We will return to the topic in 2026 with further analysis on developments shaping the use of AI in legal practice as the technological and legal landscapes continue to evolve.
Footnotes
[1] See our previous instalments
25 September 2025 Part 1 - AI for legal professionals: Where to start? |
09 October 2025 Part 2 - AI for legal professionals: Hallucinations |
23 October 2025 Part 3 - AI for legal professionals: Document review and disclosure |
06 November 2025 Part 4 - AI for legal professionals: Litigation strategy |
[2] More information on predictive analytics can be found in the fourth edition in our AI series Part 4 - AI for legal professionals: Litigation strategy |
[3] This is to be contrasted with the International Centre for Dispute Resolution, the international division of the American Arbitration Association, which has announced that AI arbitrators will be available from November 2025 for eligible document-only construction disputes.
About Hausfeld’s Digital Markets expertise
The Hausfeld Digital Markets Hub