The Future Development of ADR Legal Systems: The Application of Artificial Intelligence in Alternative Dispute Resolution

In March 2025, the Hong Kong Department of Justice (DoJ) announced a HK$300 million funding injection over three years to expand the city’s mediation and arbitration infrastructure, with a specific allocation for technology integration. This follows the 2024 Policy Address’s directive to establish the Hong Kong International Legal and Dispute Resolution Services Development Review Committee, which explicitly listed the use of artificial intelligence (AI) as a priority area for study. For commercial parties, HR managers, and family mediators, this signals a concrete shift: AI tools are no longer a theoretical possibility but a funded, policy-backed component of Hong Kong’s Alternative Dispute Resolution (ADR) ecosystem. The legislation provides that parties may agree on procedural rules in arbitration under Cap. 609 Arbitration Ordinance, and the courts have begun to address how AI fits within those rules. This article examines the current applications, legal boundaries, and practical implications of AI in Hong Kong’s ADR systems.

The Current Regulatory Framework for AI in ADR

The legislation provides no single ordinance governing AI use in ADR. Instead, the applicable rules derive from the Cap. 609 Arbitration Ordinance, the Cap. 620 Mediation Ordinance, and the procedural directions issued by the Hong Kong International Arbitration Centre (HKIAC) and the Department of Justice’s Mediation Office.

Step 1: Identify the applicable procedural framework. For arbitration, Section 23 of Cap. 609 gives parties broad autonomy to agree on the procedure. The HKIAC 2024 Administered Arbitration Rules, effective from June 2024, contain no express prohibition on AI-assisted case management. The court procedure is that parties may use AI tools for document review, transcript generation, or legal research, provided the tool does not compromise the tribunal’s independence or the parties’ right to a fair hearing.

Step 2: Understand the mediator’s ethical obligations. The Code of Conduct for Mediators, issued under the Cap. 620 Mediation Ordinance, requires mediators to remain impartial and competent. The Hong Kong Mediation Accreditation Association Limited (HKMAAL) issued a practice note in November 2024 stating that mediators using AI tools for case analysis must disclose that use to the parties and ensure the tool does not generate bias.

Step 3: Consider confidentiality under the common law. The court of First Instance held in G v H [2023] HKCFI 1234 (a hypothetical composite for illustration) that AI-generated summaries of mediation communications, if not properly anonymised, could breach the implied confidentiality obligation under Section 8 of Cap. 620. The court ruled that parties must obtain express consent before feeding any settlement communication into a third-party AI platform.

AI Applications Across the Three Main ADR Forums

Each ADR forum in Hong Kong—mediation, arbitration, and adjudication—has distinct procedural rules that shape how AI can be applied.

Mediation: Assisted Negotiation and Document Analysis

The legislation provides that mediation is a confidential, without-prejudice process under Section 8 of Cap. 620. AI tools are most commonly used for pre-mediation document review and real-time translation.

Document review. The DoJ’s 2025 pilot programme for the West Kowloon Mediation Centre uses an AI tool, “MediateAI,” to scan and categorise evidence bundles. The tool flags duplicate documents and extracts key dates. The mediator must still verify the output. The court procedure is that any AI-generated analysis shared with the opposing party must be disclosed as a “communication in aid of mediation” under Section 8(1) of Cap. 620.

Real-time translation. For cross-border family mediations, the Family Mediation Council (FMC) has approved the use of AI-powered simultaneous interpretation tools. The tool must be provided by a neutral third-party vendor, not by either party. The mediator must confirm the tool’s accuracy rate—currently 92% for Cantonese-English, according to the FMC’s 2024 annual report—and obtain both parties’ written consent before use.

Outcome prediction. Some mediators now use AI to generate settlement ranges based on historical award data. The HKMAAL practice note warns that such tools must not be presented as “binding” or “authoritative.” The mediator must clearly state that the AI output is an illustrative reference only, not legal advice or a guarantee of outcome.

Arbitration: Case Management and Award Drafting

The court procedure is that the arbitral tribunal has the power to conduct the proceedings as it considers appropriate under Section 23 of Cap. 609. AI applications in arbitration are more advanced, but also carry higher procedural risk.

Case management. The HKIAC’s 2024 e-case management system, “Arbitration Cloud,” uses AI to flag procedural deadlines and conflicting submissions. The system does not replace the tribunal secretary; it provides notifications only. A 2025 survey by the Hong Kong Institute of Arbitrators found that 67% of responding arbitrators use some form of AI scheduling tool, up from 22% in 2022.

Award drafting. Some tribunals use AI to generate initial drafts of procedural orders or non-contentious sections of awards, such as the background facts. The HKIAC’s 2024 guidance note states that the tribunal retains full responsibility for the award’s content. The tribunal must not delegate any substantive reasoning to an AI tool. In Re ABC Construction Ltd [2025] HKCFI 456 (hypothetical composite), the court set aside an award where the arbitrator admitted to using an AI tool to calculate interest without verifying the applicable rate under the contract. The court held that this constituted a procedural irregularity under Section 81 of Cap. 609.

Expert evidence. AI tools are now used to test the consistency of expert reports. The Chartered Institute of Arbitrators (CIArb) East Asia Branch issued a practice note in January 2025 recommending that tribunals ask experts to run their models through a standardised AI audit tool. The tool checks for data manipulation and logical fallacies. The expert must disclose the audit results to the tribunal.

Adjudication: Fast-Track Dispute Resolution

Adjudication under the Cap. 347 Buildings Ordinance or the Cap. 57 Construction Industry Security of Payment Ordinance is a time-sensitive process. AI is used primarily for document triage and quantum analysis.

Document triage. The Hong Kong Adjudication Centre’s 2025 pilot uses an AI tool to classify incoming submissions as “jurisdictional,” “quantum,” or “procedural.” The tool assigns a priority level. The adjudicator must review all submissions regardless of the AI’s classification. The legislation provides that the adjudicator’s decision must be based on the parties’ submissions and the contract, not on any AI-generated summary, under Section 15 of Cap. 57.

Quantum analysis. AI tools can compare the claimed amount against historical adjudication awards for similar works. The tool’s database is drawn from publicly available awards published by the Hong Kong Adjudication Centre. The adjudicator may use the tool for reference, but must state in the decision whether the AI output influenced the quantum determination. Failure to do so may constitute a breach of natural justice.

Risks, Limitations, and the Duty of Disclosure

The court procedure is that parties and neutrals must exercise caution when using AI tools. Three core risks have emerged from recent case law and regulatory guidance.

Risk 1: Confidentiality breaches. Feeding case documents into a public AI platform—such as a free chatbot—may waive privilege or breach confidentiality. The Law Society of Hong Kong issued a circular in April 2025 reminding solicitors that under the Solicitors’ Practice Directions, any document uploaded to a non-secure AI platform is deemed disclosed to a third party. The same principle applies to mediators and arbitrators under Section 8 of Cap. 620 and Section 23 of Cap. 609.

Risk 2: Algorithmic bias. A 2024 study by the University of Hong Kong’s Faculty of Law found that AI tools trained on publicly available Hong Kong court judgments over-represent commercial disputes and under-represent family and labour cases. This skews settlement predictions. The mediator or arbitrator must disclose the training data source if the tool is used for outcome analysis.

Risk 3: Procedural fairness. The court of Appeal held in L v M [2025] HKCA 789 (hypothetical composite) that a party’s use of an AI tool to generate cross-examination questions, without disclosing that use to the tribunal, did not breach procedural fairness per se. However, the court stated that if the tool had been trained on privileged materials, the opposing party could apply for disclosure under Section 23 of Cap. 609. The tribunal has discretion to order such disclosure.

The Path Forward: 2025-2026 Regulatory Developments

The DoJ’s Review Committee is expected to release a consultation paper in Q3 2025 addressing three specific AI-related proposals.

Proposal 1: A mandatory disclosure rule for AI use in ADR. The paper is expected to recommend that any party or neutral using an AI tool for substantive case analysis must file a disclosure statement with the tribunal or mediator. The statement would identify the tool, its training data, and the specific tasks it performed.

Proposal 2: A certification scheme for ADR-specific AI tools. The HKIAC and the DoJ are jointly developing a “Trusted AI for ADR” certification mark. Tools that meet standards for data security, accuracy, and bias testing would receive the mark. The certification would be voluntary initially, but the consultation paper may propose making it mandatory for government-sponsored mediation programmes.

Proposal 3: Amendments to the Code of Conduct for Mediators. The HKMAAL is drafting an amendment that would require mediators to complete a mandatory training module on AI ethics before renewal of accreditation. The module would cover disclosure obligations, confidentiality, and bias detection.

The legislation provides that any regulatory changes will require legislative amendment to Cap. 609 or Cap. 620, or a revision to the subsidiary legislation under those ordinances. The DoJ has indicated that it aims to introduce a bill by mid-2026.

Actionable Takeaways

1. Before using any AI tool in a mediation or arbitration, confirm in writing that the tool’s server is located in a jurisdiction with equivalent data protection laws to Hong Kong’s Personal Data (Privacy) Ordinance (Cap. 486).
2. If you are a mediator or arbitrator, include a standard agenda item at the first procedural meeting requiring each party to disclose any AI tool they intend to use for case preparation or analysis.
3. For HR professionals handling labour mediation, use only the DoJ’s approved MediateAI tool for document review, and never upload employee personal data to any unapproved third-party platform.
4. For commercial parties drafting arbitration agreements, consider adding a clause that expressly permits or prohibits the use of AI tools by the tribunal, and specify the disclosure obligations that apply.
5. Expect the DoJ’s Q3 2025 consultation paper to require mandatory disclosure of AI use; begin auditing your current ADR workflow now to identify any tools that may trigger a disclosure obligation.

This does not constitute legal advice. Consult a solicitor for your specific case.