Ciarb’s Guideline on the Use of AI in Arbitration is one of the most practically important documents yet produced on AI use in arbitration.
That is true not because it is dramatic, but because it is operational.
Where some institutional materials stay at the level of high-level ethics, Ciarb goes further. It discusses party use, arbitrator use, tribunal powers, disclosure, risks to fairness and enforceability, and even provides model language in the form of a template agreement and a template procedural order.
For anyone trying to understand where the field is going, that makes the guideline essential.
What the guideline is
Ciarb launched the Guideline on the Use of AI in Arbitration on September 5, 2025 and the published document is marked as updated in September 2025.
The launch note says the guideline is designed to help dispute resolvers, parties, their representatives, and other participants take advantage of AI’s benefits while mitigating risks to procedural integrity, party rights, and the enforceability of any resulting award or settlement.
That framing matters.
It means the document is not written only for arbitrators. It is written for the whole arbitral ecosystem.
Why the guideline matters
The Ciarb guideline matters for three main reasons.
1. It is broader than many competing materials
AAA-ICDR’s March 2025 guidance is important, but it is focused specifically on arbitrators’ use of AI tools. Ciarb goes further. It addresses the use of AI by parties, the powers of arbitrators to regulate that use, and the arbitrators’ own responsibilities when using AI.
2. It is more procedural
The guideline does not stop at principles. It addresses directions, rulings, model agreements, and template procedural-order language.
3. It is openly human-centered
The guideline repeatedly returns to risks involving procedural integrity, party rights, confidentiality, fairness, and enforceability. That makes it a serious governance document, not a technology celebration.
How the guideline is structured
According to the official launch note, the guideline includes:
- Part I on benefits and risks of AI in arbitration,
- Part II on general recommendations for AI use in an arbitration,
- Part III on arbitrators’ powers to give directions and make rulings on party use of AI,
- Part IV on the use of AI by arbitrators,
- Appendix A with a template agreement on the use of AI in arbitration,
- and Appendix B with a template procedural order on the use of AI in arbitration.
That structure alone shows why the document is so useful. It is not just asking whether AI use is permitted. It is asking how a proceeding should be governed if AI use is real.
The central themes
Benefits and risks
Ciarb recognizes that AI tools may help with efficiency, research, drafting, and information handling. But it pairs those advantages with a sustained discussion of risk.
The key risks are not only technical error. They include:
- threats to procedural fairness,
- confidentiality exposure,
- unequal tactical advantage,
- hidden reliance,
- and risks to the enforceability of the final outcome.
That is the right frame.
Party use of AI
One of the guideline’s most important features is that it takes party use seriously.
The document recognizes that counsel, experts, and parties may use AI in ways that affect the process. That means AI governance in arbitration cannot be reduced to the private behavior of the arbitrator alone.
Tribunal power to regulate use
Part III is especially important because it addresses the tribunal’s power to give directions and make rulings about AI use by parties.
This is a major practical step. It means the guideline is not only descriptive. It gives arbitrators a framework for active case management.
Arbitrator use of AI
Part IV addresses arbitrators’ own use of AI and reinforces the idea that responsibility remains human, not automated.
That principle has become a common institutional theme, but Ciarb situates it within a broader procedural structure rather than leaving it as a simple ethical slogan.
Why the model agreement and procedural order matter
Appendix A and Appendix B are not decorative.
They are one of the most valuable features of the guideline because they make the subject usable.
Many disputes over AI use in arbitration are not philosophical. They are practical:
- Should AI use be disclosed?
- Can confidential case materials be entered into external tools?
- Can the tribunal restrict or condition AI use?
- How should the parties address disagreement?
Model language gives parties and tribunals a starting point instead of forcing them to improvise under pressure.
That alone makes the guideline worth reading carefully.
How Ciarb differs from AAA-ICDR
Both documents are important, but they do different work.
AAA-ICDR’s March 2025 guidance is shorter and more ethics-centered around the arbitrator: accuracy, fairness, independent decision-making, transparency where material, and confidentiality.
Ciarb’s September 2025 guideline is broader and more procedural. It addresses:
- parties,
- counsel,
- arbitrators,
- tribunal powers,
- and template language for governing AI use.
If AAA-ICDR’s document tells arbitrators how to behave responsibly, Ciarb’s guideline helps the whole proceeding behave responsibly.
Why the guideline is likely to matter beyond arbitration
Ciarb’s treatment of AI is significant not only for arbitration specialists.
It also offers a framework for anyone thinking about:
- confidentiality in AI-assisted proceedings,
- disclosure thresholds,
- the line between assistance and delegation,
- and how to preserve human accountability when AI tools become common in professional practice.
The guideline’s influence may therefore reach beyond strictly arbitral settings.
Practical takeaways
For parties and counsel:
- address AI use early rather than waiting for conflict,
- decide whether disclosure expectations are needed,
- consider whether a procedural order would help,
- and treat confidentiality as a live process question.
For arbitrators:
- do not treat AI use as merely personal workflow preference,
- regulate party use where fairness or integrity is at stake,
- and remember that efficiency never displaces responsibility.
For drafters:
- the guideline’s model language is a resource,
- especially when the dispute is likely to involve sensitive information or sophisticated users of AI tools.
FAQ
When did Ciarb publish its AI guideline?
It launched the guideline on September 5, 2025, and the published document is marked as updated in September 2025.
Why is the Ciarb guideline important?
Because it is one of the most detailed institutional frameworks for AI use in arbitration and includes practical model language, not just high-level principles.
Does the guideline apply only to arbitrators?
No. One of its strengths is that it addresses parties, representatives, arbitrators, and tribunal powers over the proceeding.
What makes it different from AAA-ICDR’s guidance?
AAA-ICDR’s March 2025 guidance is narrower and focused on arbitrators. Ciarb’s guideline is broader and more procedural.
What is the most useful part of the guideline?
For many readers, the template agreement and template procedural order will be especially useful because they make the topic operational.
Conclusion
Ciarb’s AI guideline matters because it treats AI in arbitration as a governance problem that can be managed, not a novelty to be admired or feared from a distance.
That is the kind of maturity this field needs, and it is why the guideline belongs near the center of any serious Sherafy treatment of AI dispute resolution.
Further Reading
- Ciarb launch note, September 5, 2025: https://www.ciarb.org/news-listing/ciarb-launches-guideline-on-the-use-of-ai-in-arbitration/
- Ciarb Guideline on the Use of AI in Arbitration, updated September 2025: https://www.ciarb.org/media/bpndtcgu/guideline-on-the-use-of-ai-in-arbitration_updated-sept-2025.pdf
- Ciarb Education and Training AI Policy, last updated February 26, 2026: https://www.ciarb.org/general/education-and-training-ai-policy/
