AI Governance and Evidence in Dispute Resolution: An Update for ADR Institutions

The European Union's AI Act (Regulation 2024/1689) treats AI systems used by ADR bodies to research facts, interpret law or apply law to facts as high-risk under Annex III, where the proceedings produce binding legal effects. The European Commission's May 2026 Draft Guidelines confirmed that arbitration falls within this scope. Compliance obligations for high-risk systems — including human oversight, log retention, technical documentation and provider due diligence — apply from December 2, 2027 for stand-alone systems and August 2, 2028 for those embedded in broader products.

In July 2026, sixteen international arbitral institutions (including the ICC, LCIA, AAA-ICDR and ICSID) wrote jointly to the Commission, arguing that their administrative role should not attract the same regulatory burdens as the adjudicative function, which rests with the tribunal. This debate over who qualifies as the regulated "ADR body" remains unresolved.

For any institution whose neutrals hear matters involving EU parties or whose awards may be enforced within the EU, the Act's extraterritorial reach makes these questions immediate rather than theoretical.

Cross-border disputes require a wider lens: the EU AI Act, CPTPP and UN initiatives sit at different levels and do not create one interchangeable rulebook. Governing law, seat, parties and enforcement may all be relevant. See the regulatory landscape.

A consistent indicator across jurisdictions is the emergence of independent, third-party verification as a statutory requirement rather than a voluntary aspiration:

• Illinois SB 315 (signed July 2026, effective January 1, 2027) requires frontier AI developers to submit to annual independent audits — the first US state law to mandate this.
• California SB 813 and AB 1405 (signed September 2026) establish criteria for Independent Verification Organizations and a public AI Auditor Registry, creating the first state-level infrastructure for qualifying who may audit AI systems.
• The UK has signaled a potential Statutory AI Bill for 2026 following the limits of the voluntary testing model. Parliament's October 2026 hearing with frontier lab developers examined whether pre-release safety testing should become legally mandatory.
• Federal proposals in the US — including the FRONTIER Act, the AI Systems Transparency Act and the Great American AI Act — each incorporate some form of licensed or independent verification requirement, though none has yet been enacted.

The common thread is that self-attestation is no longer considered sufficient. The question shifting to ADR institutions is whether their own use of AI tools — in case management, research, document review or scheduling — may eventually fall under one or more of these frameworks, depending on the nature and impact of the tool.

Large organizations subject to the strictest requirements can carry those expectations into procurement, supply chains and partner agreements. This is the market power principle: rules may influence counterparties beyond the territory in which they were enacted.

For ADR, the relevant indicators are client due diligence, technology procurement, the terms on which information is handled, and the evidence expected of an AI-assisted process. The point is not that every institution faces the same obligations, but that commercial expectations may move before a directly applicable legal requirement does.

Fraud, incident exposure, litigation, procurement and insurance all reinforce the financial case for accountable AI. The question is broader than regulatory penalties: can an organization explain its controls when a client, insurer or decision-maker asks?

Structured evidence may support that response, but it does not guarantee lower costs, coverage or a favorable legal outcome.

What constitutes adequate evidence of an AI-assisted process is becoming more defined across multiple regulatory tracks:

• Content provenance (the C2PA standard for cryptographically sealed origin records) is now accepted in some US federal courts under Rule 901 and is integrated into platforms including LinkedIn, YouTube and TikTok.
• Record retention is specified in several statutes: Colorado's SB 26-189 requires three years of compliance records; the EU AI Act mandates at least six months of system logs for high-risk applications.
• Explainability — the ability to describe why a system produced a particular outcome — features in Colorado's consumer-facing disclosure requirements and in the EU's high-risk transparency obligations.

For ADR institutions, the implication is that the standard of evidence for AI-assisted processes will increasingly be set by statute rather than by institutional convention. Organizations that already maintain structured records of how their AI tools operate — which system ran, what it produced, and who reviewed the output — will be better positioned as these standards take effect.

The most recent legislative activity places increasing emphasis on the individual affected by an AI-assisted decision:

• Colorado's SB 26-189 gives consumers notice, explanation, correction and meaningful human review within 30 days of an adverse outcome.
• California's SB 947 gives workers notice and human corroboration for AI-involved employment decisions (effective July 1, 2027).
• The EU AI Act's Article 50 transparency obligations — requiring parties to be informed when they are interacting with an AI system — have applied since August 2, 2026.

For dispute resolution institutions, this trend intersects directly with their core function. The ability to offer a structured, credible route for challenging an AI-influenced outcome is not just a regulatory requirement — it is a service that these institutions are uniquely equipped to provide.

Accountability becomes easier to adopt when its purpose is visible to the people expected to use it. The preceding factors point to practical incentives: clearer procurement decisions, less uncertainty about responsibility, usable evidence when an outcome is challenged, and greater confidence in human review.

For ADR institutions, the question is not simply whether AI can make a process faster. It is whether the process remains understandable, contestable and accessible. Knowledge sharing can reduce duplicated effort; collaboration can help smaller organizations participate without requiring them to surrender local judgment.

This seventh factor is an analytical extension of the framework, not a claim that financial returns or regulatory approval are guaranteed. Meaningful measures would connect adoption to access, quality of explanation, timeliness of review and the experience of the people affected. That leads to the shared-access agenda considered in the conclusion.

The themes are connected: cross-border rules, national action, market expectations, financial exposure, evidence and individual recourse all shape the conditions in which AI is adopted. Institutions that track their AI use, maintain structured records and preserve human judgment will be better positioned to respond.

The purpose is not compliance in isolation. It is a more credible basis for fair and accessible dispute resolution.