13. Current international development: IMO's MASS Code

A significant development for this topic is the International Maritime Organization's adoption of the International Code of Safety for Maritime Autonomous Surface Ships (MASS Code) in May 2026. The non-mandatory Code took effect on 1 July 2026. It establishes a goal-based safety framework for covered autonomous and remotely operated cargo ships, while preserving the applicability of SOLAS and other relevant mandatory maritime instruments.

International Maritime Organization

+1

The Code is an important step, but it should not be mistaken for a complete international civil-liability code for AI-operated vessels. Its principal focus is safety and operational regulation; questions about compensation, negligence, product defects, insurance, and allocation of loss may still require analysis under other international conventions and domestic law.

The IMO's earlier regulatory scoping exercise identified several issues requiring clarification, including the meaning of “master,” “crew,” and “responsible person,” and the responsibilities associated with remote operations centres.

International Maritime Organization

+1

For a comparative maritime-governance framework, the distinction between technical safety regulation and civil liability is therefore essential.

14. Additional maritime case laws relevant to AI governance

Unlike AI-generated legal citations, maritime AI governance does not yet have a large body of reported judgments specifically deciding liability for autonomous-ship navigation. The following cases establish traditional maritime and civil-liability principles that courts may apply to disputes involving AI-enabled vessels.

The distinction is important: these are established maritime cases, not precedents holding that an AI system itself is legally liable. Their relevance lies in how existing doctrines may apply when AI is introduced into shipping operations.

Case 1. The Wagon Mound (No. 1)

Privy Council, 1961 · Overseas Tankship (UK) Ltd v Morts Dock & Engineering Co Ltd 19611961 AC 388

Facts: Oil spilled from a vessel into Sydney Harbour and spread across the water. The oil later caught fire, causing damage to a nearby wharf.

Legal issue: What standard determines whether damage caused by negligent conduct is legally recoverable?

Decision: The Privy Council held that liability in negligence depends on whether the relevant type of damage was reasonably foreseeable.

Legal principle: Foreseeability of the kind of damage is central to negligence liability.

Application to maritime AI: Suppose an autonomous vessel's navigation system ignores a foreseeable risk of collision, resulting in damage to another vessel or a port facility. A court may need to examine whether the harm was reasonably foreseeable and whether the shipowner, operator, software developer, or another party failed to take reasonable precautions.

The case does not resolve the attribution of fault among these actors. That question requires an examination of applicable maritime law, the system's design, the operational arrangements, and causation.

Case 2. The Oropesa

Court of Appeal, England, 1943 · 19431943 P 32

Facts: A maritime collision gave rise to a dispute about the consequences of a ship's actions in response to an emergency.

Legal issue: How should courts assess a ship's conduct and subsequent events when determining maritime responsibility?

Decision and principle: The case is associated with the assessment of causation in maritime incidents and the need to examine the sequence of events rather than treating every subsequent consequence as automatically attributable to the original incident.

Application to maritime AI: Autonomous navigation may involve a chain of decisions: sensor detection, risk classification, collision-avoidance manoeuvres, remote intervention, and emergency response.

A proper liability inquiry should reconstruct that chain. It should distinguish the original hazard from the software response, communication failure, remote operator's intervention, and any subsequent conduct that contributed to the loss.

The case provides a useful analytical starting point, although the specific allocation of liability for AI-related failures remains dependent on the applicable law and facts.

Case 3. The Eurymedon

Privy Council, 1974 · New Zealand Shipping Co Ltd v A.M. Satterthwaite & Co Ltd 19751975 AC 154

Facts: The dispute concerned damage occurring during cargo-handling operations and whether a third party involved in performing contractual services could rely on a limitation or exclusion clause.

Legal issue: Under what circumstances can contractual protections extend to a party that is not itself a contracting party?

Decision: The Privy Council recognised a contractual mechanism through which the stevedores could rely on the relevant protection.

Legal principle: Contractual allocation of risk, third-party protection, and the legal structure of commercial transactions can significantly influence maritime liability.

Application to maritime AI: Autonomous shipping commonly involves several contractual relationships: shipowners, charterers, operators, remote-control centres, software providers, equipment manufacturers, and maintenance contractors.

Contracts may allocate responsibility for navigation, cybersecurity, software updates, system failures, and incident reporting. However, a contractual clause does not automatically defeat the rights of an injured third party or override mandatory legal obligations.

This case is particularly useful when analysing how maritime businesses might allocate AI-related risks through contracts and insurance.

Case 4. The Nicholas H

House of Lords, 1995 · Marc Rich & Co AG v Bishop Rock Marine Co Ltd 19951995 2 AC 501

Facts: A ship classification society was alleged to have been negligent in connection with the certification of a vessel that subsequently suffered a casualty.

Legal issue: When should a classification society owe a duty of care to a party suffering economic loss because of allegedly negligent classification work?

Decision: The House of Lords rejected the relevant negligence claim, taking account of the circumstances and the policy considerations surrounding the classification society's role.

Legal principle: The existence and scope of a duty of care depend on the relationship between the parties, the nature of the function performed, and relevant legal policy considerations.

Application to maritime AI: A classification society or certification body may be involved in assessing autonomous vessel systems, safety management, or technical compliance.

If a system fails and a third party suffers loss, the mere fact that the technology was certified does not automatically establish that the certifier is liable. Conversely, the scope of any duty must be assessed under the applicable legal framework.

This case illustrates why maritime AI governance needs clearly defined responsibilities for design, certification, deployment, and continuing monitoring.

Case 5. Exxon Shipping Co. v. Baker

Supreme Court of the United States, 2008 · 554 U.S. 471

Facts: The Exxon Valdez oil spill caused extensive environmental damage. The litigation included a dispute about punitive damages under maritime law.

Legal issue: How should punitive damages be assessed in a maritime tort action?

Decision: The Supreme Court established a limitation on punitive damages in the circumstances governed by federal maritime common law.

Legal principle: Maritime liability may include remedies beyond ordinary compensatory damages, but the availability and calculation of such remedies depend on the governing law.

Application to maritime AI: Suppose an operator deploys an autonomous vessel despite known and serious navigation-system defects. A resulting collision or pollution incident could lead to claims for compensatory damages and, where legally available, additional remedies.

The relevant inquiry would include the operator's knowledge, the nature of the misconduct, causation, and the governing liability regime. Exxon Shipping does not establish that deploying AI itself justifies punitive damages.

Case 6. The CMA CGM Libra

Supreme Court of the United Kingdom, 2021 · 20212021 UKSC 51

Facts: The container ship CMA CGM Libra grounded after leaving port. The litigation concerned whether the defective passage plan rendered the vessel unseaworthy and affected the shipowner's ability to rely on contractual protections.

Legal issue: Can a deficient navigational passage plan constitute unseaworthiness?

Decision: The UK Supreme Court held that the defective passage plan rendered the vessel unseaworthy in the circumstances and that the relevant obligation of due diligence had not been satisfied.

Legal principles:

Navigational planning is part of the vessel's seaworthiness assessment.

The competence and work of those preparing the passage plan can have legal consequences.

Compliance with the relevant seaworthiness obligations is not satisfied merely because a vessel is physically operational.

Application to maritime AI: An AI navigation system may produce a route plan based on charts, weather, traffic, depth, and operational restrictions. If the plan is materially defective, the legal inquiry may extend beyond whether the software functioned as programmed.

Courts may need to examine the quality of input data, chart updates, validation, human review, and compliance with applicable navigational requirements.

This is a strong modern maritime analogy for AI-assisted passage planning, but it is not a ruling on autonomous navigation software.

Case 7. M.V. Elisabeth v. Harwan Investment & Trading Pvt. Ltd.

Supreme Court of India, 1992 · 1993 Supp (2) SCC 433

Facts: The dispute arose from maritime claims involving a foreign vessel and questions about the jurisdiction of Indian courts to exercise admiralty jurisdiction.

Legal issue: What is the scope of Indian courts' admiralty jurisdiction, including their authority in relation to foreign vessels and maritime claims?

Decision: The Supreme Court recognised the broad role of Indian admiralty jurisdiction and the importance of international maritime law principles in developing and exercising that jurisdiction.

Legal principles:

Indian admiralty jurisdiction is not confined to an unduly narrow historical conception.

Maritime claims must be considered within the relevant statutory, procedural, and international-law framework.

The existence of a foreign vessel does not, by itself, exclude Indian judicial jurisdiction where the applicable legal requirements are met.

Application to maritime AI: A collision involving an autonomous vessel may involve a foreign shipowner, an overseas remote operations centre, software developed in another country, and damage occurring in Indian waters.

The case is relevant to questions of forum, jurisdiction, and the legal mechanisms through which maritime claims may be brought in India. It does not determine which party is liable for an AI system's error; that question requires separate analysis.

Case 8. Liverpool & London S.P. & I Association Ltd v. M.V. Sea Success I

Supreme Court of India, 2003 · (2004) 9 SCC 512

Facts: The case concerned admiralty proceedings, maritime claims, and the treatment of a protection and indemnity association in the context of Indian maritime jurisdiction.

Legal issue: How should Indian courts interpret and apply admiralty jurisdiction in relation to maritime claims and the relevant parties?

Decision and principle: The Supreme Court examined the scope of admiralty jurisdiction and the relationship between the maritime claim and the legal entities involved.

Application to maritime AI: In a casualty involving an AI-enabled ship, claimants may need to identify the appropriate defendant, available maritime remedies, and relevant insurance arrangements. The involvement of an AI supplier or remote operator may complicate the identification of responsible parties.

This case is useful for understanding the jurisdictional and institutional dimensions of maritime litigation, although it does not directly decide liability for autonomous vessels.

15. Comparative maritime AI governance frameworks

The governance of AI-enabled maritime operations is different from the governance of AI in ordinary civil litigation. It involves international navigation, flag-state jurisdiction, coastal-state authority, port regulation, safety certification, marine environmental protection, insurance, and private-law liability.

The following comparison identifies the main regulatory approaches.

Checking Favorites status

A. International framework — IMO and UNCLOS

The International Maritime Organization provides the principal international institutional framework for ship safety and pollution prevention. The United Nations Convention on the Law of the Sea governs matters including navigation, flag-state jurisdiction, and coastal-state rights and obligations.

The 2026 MASS Code establishes a goal-based safety framework for covered autonomous cargo ships. It supplements rather than replaces applicable international maritime instruments.

International Maritime Organization

+1

Strength: International coordination.

Challenge: International rules must address different levels of autonomy and clarify responsibility across borders.

Checking Favorites status

B. India — Admiralty, statutory safety, and environmental regulation

India's framework draws on the Admiralty (Jurisdiction and Settlement of Maritime Claims) Act, 2017, the Merchant Shipping Act, 1958, as amended, applicable port legislation, and relevant international obligations.

The Indian approach must reconcile the benefits of maritime automation with ship safety, seafarer protection, pollution prevention, and effective remedies for injured parties.

Strength: Existing maritime jurisdiction and established civil remedies.

Challenge: Allocation of responsibility where the owner, remote operator, software developer, and equipment supplier are located in different jurisdictions.

Checking Favorites status

C. United Kingdom — Safety regulation and maritime common law

The UK combines international maritime obligations, domestic ship-safety legislation, maritime common law, and established negligence and contractual principles.

Cases such as The CMA CGM Libra demonstrate the continuing importance of seaworthiness and proper navigational planning.

Strength: Established doctrines for assessing negligence, seaworthiness, and contractual allocation of risk.

Challenge: Applying traditional standards to software failures, remote operations centres, and autonomous navigation.

Checking Favorites status

D. United States — Federal maritime law and risk allocation

US maritime governance combines federal maritime law, applicable federal statutes, navigation requirements, environmental legislation, and established doctrines of negligence and comparative fault.

Exxon Shipping Co. v. Baker illustrates the role of maritime common law in determining remedies.

Strength: Developed maritime liability principles and established litigation mechanisms.

Challenge: Existing doctrines may not fully resolve how fault should be divided among operators, manufacturers, and software providers in autonomous-ship casualties.

Checking Favorites status

E. European Union — Risk regulation and digital governance

EU maritime governance operates alongside international maritime conventions and applicable EU legislation. The EU AI Act may apply to relevant AI systems depending on their intended purpose and classification, but it should not be assumed that every ship-navigation AI system automatically falls within a particular high-risk category.

Strength: Structured attention to system risk, governance, and accountability.

Challenge: Coordinating AI-specific obligations with maritime safety rules, flag-state responsibilities, and international liability conventions.

China, Norway, Japan, and South Korea also provide important comparative contexts because of their maritime industries, shipbuilding capabilities, and autonomous-shipping research. Their individual domestic rules and trial-approval requirements must be examined separately rather than inferred solely from their participation in international maritime institutions.

16. Core international legal instruments

A detailed academic analysis should consider the interaction of the following instruments:

The applicability of each instrument depends on matters such as the vessel, the type of casualty, the parties, the relevant state obligations, and the convention's scope and status.

The IMO's regulatory work has specifically identified issues involving remote operators, the role of the master, operational safety, and the application of existing instruments to different degrees of autonomy.

International Maritime Organization

+2

17. Civil liability for AI-related maritime accidents

Civil liability is one of the most difficult parts of maritime AI governance because several different causes may contribute to a single casualty.

Consider an autonomous cargo vessel that collides with a fishing vessel after its perception system incorrectly classifies the fishing vessel's movement.

Potentially relevant questions include:

Did the shipowner maintain the vessel and its systems properly?

Did the operator establish appropriate operating limits?

Did the navigation software contain a design or implementation defect?

Were the sensors or navigational data unreliable?

Did a remote operator fail to intervene when intervention was required?

Did the vessel comply with the applicable collision-avoidance rules?

Did a cybersecurity incident contribute to the collision?

Which party's conduct legally caused the damage?

A sound liability analysis should separate the following issues.

17.1 Negligence

The claimant may need to establish the relevant duty of care, breach, causation, and damage under the applicable law. The standard of care may include reasonable measures for testing, maintaining, monitoring, and safely operating the system.

17.2 Product liability

Where a defective navigation system or other component causes injury or property damage, product-liability legislation or other applicable legal doctrines may be relevant.

Whether the software constitutes a product, whether a defect exists, and which party bears responsibility depend on the applicable legal framework.

17.3 Contractual liability

Contracts between owners, operators, charterers, equipment manufacturers, and service providers may allocate certain risks and obligations.

Such arrangements must be distinguished from the rights of third parties and any mandatory statutory or convention-based obligations.

17.4 Maritime limitation of liability

Applicable maritime law may permit certain parties to limit liability for qualifying claims, subject to the relevant convention or domestic regime and any applicable exceptions.

The fact that an AI system contributed to an accident does not automatically remove or establish a right to limitation.

17.5 Insurance and compensation

Marine insurance, protection and indemnity arrangements, and any applicable compulsory insurance regime may determine how a covered loss is financed.

A separate question is whether the insurer can recover from another responsible party through subrogation or another legally recognised mechanism.

18. Proposed international governance framework for autonomous ships

An effective framework should not treat AI as a legal person that absorbs responsibility for an accident. It should identify the human and corporate actors responsible for designing, operating, maintaining, certifying, and supervising the system.

The following is a proposed framework for future development.

Layer 1 — Safety certification

Require risk assessment, validation, operational limits, redundancy, fail-safe behaviour, and appropriate certification for the vessel's level of autonomy.

Layer 2 — Human accountability

Identify the master, remote operator, company responsible for operation, and the individuals authorised to intervene in an emergency.

Layer 3 — Technical and cyber governance

Require secure communications, software-change management, logging, incident reporting, and controls against unauthorised access.

Layer 4 — Civil liability and compensation

Clarify the interaction of shipowner liability, operator negligence, product liability, contractual allocation, insurance, and maritime limitation rules.

Layer 5 — Jurisdiction and evidence

Establish workable procedures for preserving voyage data, accessing relevant technical evidence, resolving cross-border jurisdictional questions, and protecting legitimate confidentiality.

Layer 6 — International coordination

Develop interoperable requirements through the IMO and coordinate them with relevant national laws and international civil-liability regimes.

A key requirement is that the system must produce enough reliable records to allow investigators and courts to reconstruct its decisions. Without adequate technical evidence, even well-developed liability rules may be difficult to apply.

19. Conclusion

Civil law and comparative AI maritime governance concern the interaction between traditional legal responsibility and increasingly autonomous technology. The challenge is not limited to whether an AI-enabled ship can navigate safely. It also concerns who is legally responsible when a system fails, how affected parties obtain compensation, which court has jurisdiction, and how safety and environmental obligations are enforced.

The eight maritime cases discussed above supply established principles concerning negligence, causation, contractual risk allocation, seaworthiness, maritime jurisdiction, and remedies. They do not constitute a body of direct AI-liability precedent, but they provide a foundation for applying existing law to new technological conditions.

The 2026 IMO MASS Code represents a major step toward international safety governance for autonomous shipping. Yet comprehensive governance also requires coordination between technical standards, flag-state and coastal-state responsibilities, civil liability, insurance, evidence preservation, and access to effective judicial remedies.

The preferred model is internationally coordinated safety regulation combined with clearly allocated civil liability and effective human oversight. Autonomous technology should improve maritime safety and operational efficiency without creating gaps in legal responsibility or leaving victims without an effective remedy.

The ultimate objective is to ensure that innovation at sea remains consistent with the principles of accountability, compensation for legally recognised harm, navigational safety, and protection of the marine environment.

Academic note: The case discussions are explanatory summaries. For formal submission, verify the judgments' full texts, exact holdings, and pinpoint references against authorised law reports. The relevance of traditional maritime cases to AI-related disputes is analytical and does not mean that those courts decided autonomous-ship liability questions.