1. Introduction

Artificial Intelligence (AI) arbitration optimization frameworks refer to the legal rules, technological systems, and procedural methods used to improve arbitration through AI-powered tools. These tools may assist in selecting arbitrators, managing evidence, predicting procedural delays, organizing legal documents, calculating damages, identifying relevant precedents, and facilitating settlement.

In Morocco, this subject lies at the intersection of civil law, contract law, arbitration, digital transformation, data protection, and access to justice.

The central legal principle is that AI may assist arbitration, but it should not eliminate the parties’ right to a fair hearing, the arbitrator’s independent judgment, or judicial supervision required by law.

Morocco's framework must be understood through its arbitration legislation, particularly Law No. 95-17 on arbitration and conventional mediation, enacted in 2022, together with the Moroccan Code of Obligations and Contracts, applicable procedural legislation, and data-protection rules.

A significant distinction is necessary: AI arbitration optimization is an emerging legal and technological field, not a separate, fully codified branch of Moroccan law. The legal analysis therefore combines Moroccan legislation with relevant comparative case law. The cases below are not presented as Moroccan judgments specifically deciding AI arbitration unless expressly identified as such.

2. Legal framework in Morocco

A. Law No. 95-17 on arbitration and conventional mediation

Law No. 95-17 establishes the modern statutory framework governing arbitration and conventional mediation in Morocco. It addresses matters such as arbitration agreements, arbitral proceedings, arbitral awards, and the relationship between arbitration and the courts.

Its relevance to AI-assisted arbitration includes the following principles:

Party autonomy: Parties may agree to arbitrate disputes and establish permissible procedural arrangements.

Arbitrator independence: The decision-maker must exercise independent judgment and comply with applicable impartiality requirements.

Procedural fairness: Each party must have a meaningful opportunity to present its case and respond to the other party’s arguments.

Judicial supervision: Arbitral awards remain subject to the forms of judicial scrutiny and challenge permitted by applicable law.

Enforceability: An award must satisfy the applicable legal requirements to receive recognition or enforcement.

AI tools may support these processes, but the use of an algorithm cannot override mandatory statutory requirements.

B. Moroccan Code of Obligations and Contracts

The Moroccan Code of Obligations and Contracts provides the general civil-law foundation for contractual relationships. Its principles are relevant when parties agree to AI-assisted dispute resolution through a commercial contract.

For example, a contract might provide that:

Disputes will be referred to institutional arbitration.

Proceedings may use electronic documents and remote hearings.

AI software may help organize evidence and calculate financial claims.

The arbitrator retains authority to evaluate evidence and determine the outcome.

The agreement should identify the scope of AI assistance clearly. A vague clause permitting an automated system to determine all disputes could raise questions about consent, the identity of the decision-maker, procedural fairness, and the enforceability of the resulting award.

C. Personal data protection

Morocco's Law No. 09-08 on the protection of individuals with regard to the processing of personal data is particularly important when arbitration platforms process confidential business records, employee information, customer data, financial transactions, or identifying information.

An AI arbitration platform should address:

Lawful grounds for processing personal data.

Appropriate security and access controls.

Restrictions on unnecessary collection and retention.

Confidentiality of arbitration documents.

Applicable requirements for transferring data outside Morocco.

Risks arising from uploading sensitive evidence to third-party AI services.

The legal analysis must also consider the nature of the information, the platform's location, and the parties' contractual and statutory obligations.

D. Electronic proceedings and remote arbitration

Law No. 95-17 also provides a basis for modern procedural arrangements. Its provisions address electronic submissions, remote meetings in specified circumstances, and electronic arbitral awards. These provisions are particularly relevant to digital arbitration optimization.

Aceris Law

+1

A practical framework could therefore combine legally compliant electronic procedures with AI-assisted document analysis, provided that the parties’ agreement, applicable arbitration rules, and statutory requirements are respected.

3. Meaning and objectives of AI arbitration optimization

AI arbitration optimization is the structured use of computational tools to improve the efficiency, accuracy, transparency, and procedural management of arbitration.

Document and evidence analysis

AI identifies relevant contracts, invoices, correspondence, duplicate documents, and inconsistencies in large evidence collections.

Procedural scheduling

Software can propose hearing dates, track deadlines, organize procedural orders, and identify bottlenecks.

Legal research and analysis

AI can retrieve relevant statutory provisions and judgments, compare legal arguments, and help identify questions requiring further investigation.

Damages and financial analysis

Models can calculate alternative damages scenarios, contractual interest, unpaid invoices, and financial losses using verified inputs.

Fairness and compliance monitoring

Oversight tools can flag missing submissions, unequal procedural treatment, confidentiality risks, or potential conflicts requiring human review.

The objective is not simply to make arbitration faster. A sound optimization framework must balance efficiency with procedural equality, confidentiality, accuracy, independence, and enforceability.

4. Core legal principles governing AI-assisted arbitration

A. Party consent and autonomy

Arbitration derives its authority from the arbitration agreement and the applicable law. Under Morocco's Law No. 95-17, parties may agree to arbitration for disputes arising from legal relationships within the statutory framework.

AI raises an additional question: what exactly have the parties consented to?

There is an important distinction between agreeing to use AI as an administrative tool and agreeing to let AI make the final decision.

Administrative assistance: AI organizes documents, manages calendars, and assists with calculations.

Analytical assistance: AI identifies arguments, summarizes evidence, or proposes possible legal interpretations for the arbitrator to assess.

Automated adjudication: A system independently determines liability, awards damages, or issues a binding decision.

The first two uses can be incorporated into an appropriate arbitration procedure with safeguards. The third raises more fundamental questions about consent, the qualifications and identity of the decision-maker, accountability, and procedural fairness.

A Moroccan arbitration clause should expressly describe any material use of AI rather than relying on an ambiguous general reference to digital technology.

B. Equality of parties and the right to be heard

An AI system must not create an unfair procedural advantage for one party.

Suppose a Moroccan construction company submits 40,000 pages of documents to an AI platform, while the opposing company has limited resources and cannot use comparable analytical tools. The arbitrator should not automatically treat the first company's AI-generated summary as more reliable.

Appropriate safeguards include:

Giving both parties an opportunity to challenge material AI-generated summaries.

Disclosing significant AI-assisted analytical methods when necessary for a fair hearing.

Providing access to the underlying evidence relevant to disputed findings.

Allowing parties to correct inaccurate translations, summaries, or extracted data.

Ensuring that procedural deadlines remain reasonable.

AI-generated outputs are not substitutes for the evidence itself.

C. Independence and impartiality

An arbitrator must not surrender independent judgment to a software provider, algorithm, party, or institution.

Potential risks include biased training data, undisclosed commercial relationships with technology vendors, conflicts of interest, and automated recommendations that systematically favor one type of claimant.

A responsible framework should require the arbitrator to evaluate relevant evidence independently and disclose conflicts where required by applicable law and arbitration rules.

D. Reasoned awards and accountability

A reasoned arbitral award should reflect the tribunal's assessment of the dispute, not merely repeat an AI-generated conclusion.

Where AI materially assists the analysis, the arbitrator should be able to explain the relevant factual findings, legal rules, and reasoning without relying on an unexplained algorithmic score.

This matters because challenges to awards may concern the validity of the arbitration agreement, the constitution of the tribunal, procedural irregularities, or public policy, depending on the applicable legal regime.

E. Confidentiality and cybersecurity

Arbitration often involves commercially sensitive information, trade secrets, financial statements, and personal data.

An AI optimization framework should establish:

Approved software and authorized users.

Encryption and access restrictions.

Rules governing the retention and deletion of documents.

Restrictions on using confidential submissions to train public or third-party models.

Procedures for reporting and responding to data breaches.

Compliance with applicable Moroccan data-protection law and cross-border transfer requirements.

Confidentiality should be designed into the system rather than assumed simply because arbitration proceedings are private.

5. Proposed AI arbitration optimization framework for Morocco

The following is a proposed governance model, not a description of an officially adopted Moroccan AI-arbitration code.

Stage 1 — Legal and contractual validation

Confirm the written arbitration agreement, applicable law, seat, institutional rules, scope of permitted AI assistance, and allocation of decision-making authority.

Stage 2 — Secure evidence intake

Verify document provenance, preserve original evidence, control access, classify confidential information, and establish lawful data-processing arrangements.

Stage 3 — AI-assisted case management

Use AI for document classification, translation, chronology building, deadline tracking, and identification of disputed issues.

Stage 4 — Human verification and adversarial testing

Require the parties and tribunal to examine material outputs, correct errors, test alternative interpretations, and preserve each party's opportunity to respond.

Stage 5 — Independent decision and review

The arbitrator independently determines the facts, applicable law, liability, and remedy, and issues an award satisfying legal and procedural requirements.

Practical performance indicators

The framework can be evaluated through measurable indicators:

These indicators should be assessed together. A platform that reduces hearing time but introduces serious errors or compromises fairness cannot be considered successful merely because it is faster.

6. At least 6 important case laws relevant to AI arbitration optimization

Important clarification: The following are genuine judicial decisions from comparative jurisdictions, not six Moroccan cases on AI arbitration. They are relevant because they address consent, impartiality, procedural fairness, enforceability, and judicial control—issues that would arise if AI were used in Moroccan arbitration.

Case 1: Dallah Real Estate and Tourism Holding Co. v. Ministry of Religious Affairs, Government of Pakistan (2010)

United Kingdom Supreme Court · [2010] UKSC 46

Facts: Dallah entered into an arrangement concerning accommodation for Pakistani pilgrims. A dispute arose, and Dallah obtained an arbitral award against Pakistan. Pakistan disputed that it was bound by the arbitration agreement.

Legal issue: Could the award be enforced against a party that had not consented to arbitration?

Decision: The UK Supreme Court refused enforcement against Pakistan because it was not a party to the arbitration agreement under the applicable law.

Relevance to AI arbitration in Morocco:

Arbitration depends on legally effective consent.

AI software cannot create an arbitration agreement merely by identifying a commercial relationship.

Automated systems must distinguish between a signatory, an authorized representative, and a third party.

Before an AI-assisted proceeding begins, the tribunal must establish the proper parties and its jurisdiction.

Principle: AI may assist with identifying contractual relationships, but it cannot replace the tribunal's legal determination of consent and jurisdiction.

Case 2: First Options of Chicago, Inc. v. Kaplan (1995)

United States Supreme Court · 514 U.S. 938

Facts: A dispute arose concerning trading losses and whether certain parties were bound by an arbitration arrangement. The parties contested who should determine the question of arbitrability.

Legal issue: When should a court defer to an arbitrator's decision about the arbitrator's own jurisdiction?

Decision: The Supreme Court held that courts generally determine arbitrability independently unless the parties clearly and unmistakably agreed to submit that question to the arbitrator.

Relevance to AI arbitration:

An AI platform should not automatically decide whether a dispute falls within an arbitration clause.

A contractual clause authorizing AI document review does not necessarily authorize AI to determine jurisdiction.

The parties' intention regarding delegated decision-making must be established under the applicable law.

Automated jurisdictional recommendations should remain subject to proper legal review.

Principle: Consent to arbitration and consent to delegate particular decisions are distinct questions.

Case 3: Halliburton Company v. Chubb Bermuda Insurance Ltd (2020)

United Kingdom Supreme Court · [2020] UKSC 48

Facts: An arbitrator accepted appointments in multiple arbitrations involving overlapping subject matter and a common party without making the necessary disclosure to one of the parties in the first arbitration.

Legal issue: Could nondisclosure of overlapping appointments give rise to justifiable doubts about an arbitrator's impartiality?

Decision: The Supreme Court confirmed the relevant duty of disclosure under English law but dismissed the application to remove the arbitrator on the facts of the case.

International Bar Association

+1

Relevance to AI arbitration:

AI optimization systems can introduce conflicts that are less obvious than traditional conflicts of interest. Examples include:

An arbitrator's undisclosed financial relationship with an AI vendor.

A case-management system provided by one of the parties.

A recommendation model developed using confidential information from related proceedings.

Repeated use of a model whose outputs systematically favor one party's interests.

Principle: Technology must not undermine independence, impartiality, or appropriate disclosure.

Case 4: Uber Technologies Inc. v. Heller (2020)

Supreme Court of Canada · 2020 SCC 16

Facts: A worker challenged an arbitration clause in an agreement with Uber. The clause required disputes to be arbitrated in the Netherlands and involved substantial costs that made the process practically inaccessible.

Legal issue: Could an arbitration clause be invalid because it was unconscionable and effectively prevented access to justice?

Decision: The Supreme Court of Canada held the arbitration clause unconscionable and invalid under the law applicable to the dispute.

Relevance to AI arbitration in Morocco:

AI may reduce document-review expenses, but a digital arbitration framework could also create new barriers through expensive subscriptions, proprietary software, language restrictions, or inaccessible digital interfaces.

For example, a small Moroccan supplier should not be effectively denied a hearing because it cannot afford the AI platform selected by a larger commercial party.

Principle: Procedural efficiency must not come at the expense of meaningful access to dispute resolution.

Case 5: Mitsubishi Motors Corp. v. Soler Chrysler-Plymouth, Inc. (1985)

United States Supreme Court · 473 U.S. 614

Facts: A commercial dispute arose from an international automobile distribution agreement. The dispute included claims involving competition law and the enforceability of an arbitration agreement.

Legal issue: Could international commercial arbitration be used to resolve statutory claims, including claims arising under competition law?

Decision: The Supreme Court generally upheld arbitration of the statutory claims at issue, emphasizing the importance of enforcing international arbitration agreements while recognizing the significance of applicable statutory rights.

Relevance to AI arbitration in Morocco:

AI systems may help tribunals analyze complex statutory claims, contractual restrictions, market evidence, or economic damages. However:

AI cannot determine that mandatory law is irrelevant merely because a contract selects another legal framework.

The arbitrator must independently examine applicable legal rules.

The use of AI must not deprive parties of substantive rights that the applicable law protects.

Principle: Arbitration can provide a mechanism for resolving complex commercial disputes, but the applicable law and legally protected rights remain important.

Case 6: Ssangyong Engineering & Construction Co. Ltd. v. National Highways Authority of India (2019)

Supreme Court of India · (2019) 15 SCC 131

Facts: The dispute concerned a construction contract, changes in the method of calculating price adjustments, and the limits of judicial interference with an arbitral award.

Legal issue: When can an arbitral award be challenged for violating public policy or exceeding the legally permissible scope of arbitral decision-making?

Decision: The Supreme Court examined the statutory grounds for setting aside arbitral awards and the limits of public-policy review, finding the award vulnerable under the applicable Indian arbitration law.

Relevance to AI arbitration in Morocco:

An AI system might recommend a damages calculation or contractual interpretation based on a statistical pattern. If the arbitrator adopts that recommendation without examining the contract or applicable law, the award could become vulnerable to challenge under whatever review grounds govern the particular arbitration.

The case illustrates why:

The arbitrator must independently interpret the contract.

AI-generated calculations must be verified.

The award must address material legal arguments.

Courts must apply the relevant statutory grounds for review rather than simply defer to an algorithm.

Principle: Automated assistance does not eliminate the need for legally reasoned and reviewable arbitral decisions.

Case 7: Centrotrade Minerals & Metal Inc. v. Hindustan Copper Ltd. (2017)

Supreme Court of India · (2017) 2 SCC 228

Facts: The dispute concerned a commercial contract providing for a two-tier arbitration mechanism, including a subsequent arbitral stage after the first award.

Legal issue: Was a two-tier arbitration arrangement legally permissible?

Decision: The Supreme Court upheld the validity of the two-tier arbitration arrangement under the applicable legal framework, while the litigation history involved additional questions concerning its operation and enforcement.

Relevance to AI arbitration:

This case offers a useful analogy for designing review procedures around technology-assisted decisions. Commercial parties may consider staged processes such as:

AI-assisted initial document analysis.

Human arbitral determination.

A further arbitral stage where the agreement and applicable law permit it.

However, an AI-generated preliminary assessment is not itself a second-tier arbitral award, and parties cannot assume that an algorithmic appeal mechanism is legally equivalent to an established arbitral procedure.

Principle: Procedural design should follow the parties' agreement and the applicable arbitration law.

Case-law comparison table

These cases are persuasive comparative material for policy development and legal analysis. Their holdings arise under their respective national laws and must not be treated as binding interpretations of Moroccan legislation.

7. Application to major sectors in Morocco

A. Construction and infrastructure disputes

Moroccan construction disputes may involve delay claims, variations, payment certificates, engineering reports, and extensive project correspondence.

AI could create chronological records, identify inconsistent progress reports, calculate alternative delay scenarios, and organize expert evidence.

Legal safeguard: The arbitrator must verify the contractual basis of the claim and assess competing expert opinions rather than adopting the AI system's conclusion automatically.

B. Banking and financial disputes

AI may assist in analyzing loan agreements, repayment histories, interest calculations, financial transactions, and documentary evidence.

The principal risks include inaccurate calculations, exposure of sensitive customer information, and reliance on models that cannot explain their outputs.

Legal safeguard: Financial calculations must be independently verifiable, and personal data must be processed in accordance with applicable legal requirements.

C. International commercial arbitration

Morocco's role in international trade makes AI-assisted arbitration potentially relevant to cross-border sales, distribution, logistics, investment, and supply-chain disputes.

A framework should identify the seat of arbitration, governing law, procedural rules, language, data location, and enforcement considerations.

A Moroccan-seated arbitration and a foreign-seated arbitration involving a Moroccan business may be governed by different procedural and enforcement rules.

D. Intellectual property and technology disputes

AI may help organize software licences, copyright ownership records, source-code evidence, and claims involving unauthorized use of confidential information.

However, the platform must protect trade secrets and should not be given unnecessary access to commercially sensitive source code.

E. Consumer and small-business disputes

AI-assisted mediation and case-management systems may help parties understand claims and evaluate settlement options.

Nevertheless, an automated settlement recommendation should not be confused with a binding arbitral award. The distinction must be clear to the parties, especially where bargaining power is unequal.

8. Major legal challenges

The principal risks of AI-assisted arbitration in Morocco can be summarized as follows.

Algorithmic bias

Historical decisions or biased training data may produce systematically inaccurate or unfair recommendations.

Fabricated authorities

Generative AI can invent case names, quotations, legal provisions, or precedents. Every cited authority should be independently verified.

Confidentiality and privacy

Uploading arbitration records to an inadequately secured external system may expose trade secrets or personal information.

Unclear accountability

A dispute may arise over responsibility for an erroneous AI-generated recommendation. Responsibility cannot simply be transferred to the software.

Language and translation

Arabic, French, and other languages used in commercial dealings may create translation and interpretation risks. Material translations should be checked by qualified reviewers.

Enforcement and judicial review

A technology-assisted award must still satisfy the requirements applicable to the arbitration, including any relevant requirements concerning tribunal composition, procedure, reasons, and public policy.

9. Recommendations for Morocco

A practical approach would be to develop AI arbitration governance through coordinated measures by legislators, arbitral institutions, arbitrators, technology providers, and professional bodies.

1. Clarify permitted AI uses. Establish clear distinctions between administrative assistance, legal analysis, and automated adjudication.

2. Require meaningful human control. The tribunal should independently determine jurisdiction, evaluate evidence, interpret the law, and decide remedies.

3. Develop procedural disclosure standards. Parties should understand material uses of AI that may affect the presentation or assessment of their cases.

4. Adopt secure data-management rules. Require appropriate safeguards for confidential documents, personal information, access controls, and external data transfers.

5. Create audit and verification procedures. Preserve source documents, record material AI-assisted steps where appropriate, and check citations, calculations, and translations.

6. Protect procedural equality. Provide both parties with a meaningful opportunity to challenge relevant AI-generated material and avoid making access to a particular commercial platform a condition of effective participation.

7. Train arbitrators and practitioners. Training should cover algorithmic limitations, evidentiary reliability, cybersecurity, privacy, and the legal implications of automated recommendations.

8. Evaluate outcomes. Institutions could assess whether AI actually reduces costs and delays without increasing procedural objections, errors, or enforcement difficulties.

10. Conclusion

Civil Law and AI Arbitration Optimization Frameworks in Morocco concern the responsible integration of AI into the existing arbitration system rather than the replacement of that system.

Law No. 95-17 provides the central statutory framework for arbitration and conventional mediation. Contract law, data-protection requirements, applicable procedural rules, and the legal requirements governing arbitral awards remain essential to any technology-assisted process.

The seven comparative cases demonstrate the legal considerations that should guide the development of AI-assisted arbitration: consent, jurisdiction, impartiality, access to justice, substantive legal rights, enforceability, and lawful procedural design.

The most defensible model for Morocco is human-led arbitration supported by carefully controlled AI tools. AI can improve document review, scheduling, translation, research, and financial analysis. However, the tribunal must retain independent decision-making authority, and parties must retain the procedural protections afforded by the applicable law.

The long-term success of this framework should be measured not only by speed and cost reduction but also by accuracy, confidentiality, fairness, transparency, and enforceability.