New guidelines address the EU AI Act and arbitrators’ use of AI
The Danish Institute of Arbitration presented its new Guidelines on the Regulation of Arbitrators’ Use of AI at Copenhagen Arbitration Days in early October 2026. They are the first guidelines of their kind to focus specifically on what the EU AI Act means for arbitrators’ use of AI.
Niels Chr. Ellegaard, PhD researcher at the University of Copenhagen, led the drafting of the guidelines. His research examines AI and human legal decision-making by courts and arbitral tribunals, with a particular focus on how responsibility should be shared between humans and AI systems. Jørgen Bek Weiss Hansen (participating in the CCBE working Group on AI) also made a substantial contribution to the guidelines together with the valuable inputs from Steffen Pihlblad (Secretary-General at the Danish Institute of Arbitration).
The guidelines help arbitrators assess when using AI as a tool, a source of information or decision support may trigger the AI Act’s high-risk rules. They explain the tribunal’s potential responsibilities and offer practical recommendations on human oversight, transparency and discussing AI use with the parties. Throughout, the arbitrators remain responsible for their decisions.
Although the relevant high-risk rules will apply from 2 December 2027, guidance is needed now: arbitration proceedings already under way may continue beyond that date. Decisions about AI use made today could therefore affect an award rendered after the rules become applicable.