Strengthening Supply Chains through Efficiency, Resilience, AI and Environmental Performance

2. The role of AI powered trade facilitation for supply chain efficiency and resilience

AI is becoming increasingly central to how international trade functions. As AI technology is rapidly advancing, it is reshaping the way goods are moved, transactions are processed, and supply chain operations are managed. AI technologies underpin logistics optimisation, predictive analytics, and real‑time visibility, enabling firms to anticipate disruptions and allocate resources more efficiently. Yet AI’s ability to enhance trade depends heavily on the digital maturity of border and supply chain processes: without structured, machine‑readable data, interoperable border management related systems, and integrated digital platforms, AI cannot deliver meaningful gains in customs or logistics operations.

These advances have important implications for trade facilitation. At the border, AI can significantly strengthen core functions such as risk profiling, targeting, anomaly detection, document processing, and the verification of harmonized system (HS) codes and certificates. Combined with robotics, IoT devices, and automation systems, AI can improve the accuracy and speed of inspections while reducing manual workloads. Beyond the border, AI can enhance transparency, traceability, and risk management across entire supply chains – extending the benefits of trade facilitation into upstream production and downstream distribution. When embedded in digital, interoperable systems, AI amplifies the ability of traders and authorities to manage growing complexities and respond to supply chain disruptions.

2.1. AI is transforming trade

Data, AI services, and the goods that enable AI technologies are increasingly traded across borders. There are three defining features of this evolving landscape in terms of what we trade today (Ferencz, López González and Oliván García, 2022[4]). First, trade in AI-enabled goods, including “smart products” combining the characteristics of goods and services, high-performance computing hardware, data centre infrastructure, and connectivity equipment (WTO, 2025[5]). Second, cross-border trade in AI services, including generative AI, is expanding. Third, cross-border data flows are rising fast, driven by the need to train, retrain and fine-tune AI models and enable real-time inference (Sun and Trefler, 2023[6]).

At the same time, AI technologies are transforming supply chain management by improving inventory control, demand forecasting, and logistics optimisation. For example, recent studies find that companies that adopted AI-driven supply chain solutions have achieved significant reduction in logistics costs (Cohen and Tang, 2024[7]). AI can also enhance supply chain visibility by detecting patterns and anomalies across vast datasets from multiple sources to track shipments and anticipate disruptions more accurately and efficiently (McKinsey & Company, 2025[8]). But, beyond supply chain management, AI also stands to revolutionise trade facilitation.

2.2. AI can revolutionise trade facilitation

Emerging digital technologies – including AI, big data analytics, and distributed ledger systems – are reshaping customs operations and the wider ecosystem of border management. These technologies are increasingly embedded in how customs administrations analyse trade flows, assess financial and tax information, and detect irregularities.

AI systems, in particular, have the potential to:

• Support more sophisticated risk segmentation and anomaly detection by drawing on large volumes of data sources (Vijayakumar, 2025[9]), allowing risk profiles to be refined more effectively than with rule‑based systems alone;

• Enable machine learning models to rapidly identify patterns associated with non‑compliance or illicit activity (World Customs Organization, 2025[10]), enabling the detection of fraud and other high‑risk behaviours across large transactions;

• Improve the consistency and timeliness of operational choices through automated decision support tools, allowing border officers to focus on other tasks (Outram and Swartz, 2026[11]); and

• Beyond risk management, AI systems can contribute to the automation of key processes, from document verification and data matching to the integrity checks required for certificates and authorisations (Anderson, Belcineanu and Tzvetkova, 2025[12]).

Together, these innovations can enhance operational efficiency, strengthen the quality of controls, and improve the overall performance and predictability of border procedures.

However, the effective use of AI for trade facilitation is not automatic and depends on two fundamental enabling conditions.

First, border processes must be sufficiently digitalised, enabling the transition to digital customs and allowing trade facilitation to go paperless (OECD, 2025_[13]) (Mikuriya, 2016[13]). AI systems rely on structured, machine‑readable, and interoperable data, which can only be generated and exchanged efficiently when documents, procedures, and information flows have been digitised across the supply chain. Without this foundation, AI cannot support automation, predictive modelling, or reliable risk analysis.

Second, the broader trade policy landscape must be conducive to the adoption and diffusion of AI technologies. This includes ensuring access to AI‑enabling goods and digital services, maintaining open and trusted cross‑border data flows, enabling the movement of skilled personnel, and establishing governance frameworks that provide legal certainty for electronic transactions (Ferencz, López González and Oliván García, 2022[4]). When these conditions are in place, countries are better positioned to harness AI as a tool for more effective, agile, and transparent trade facilitation.

These dimensions are discussed in greater detail below.

2.2.1. Going paperless is an essential pre-condition to reap the full benefits of AI

AI systems rely on complete, high‑quality, and trustworthy digital inputs. Paperless trade – built on digital documents, standardised data formats, and interoperable processing systems – provides the essential infrastructure on which AI can function effectively. Without these foundations, AI cannot automate classification tasks, link documents across different stages of a transaction, or generate predictive insights that support supply chain decision‑making. Digitalisation of trade facilitation frameworks must also be backed by robust legal and regulatory frameworks: frameworks for electronic transactions, e‑authentication, e‑signatures, and cross‑border data exchange all help create the certainty needed for the seamless movement and use of digital information (OECD, 2025[14]). Even with such systems in place, however, effective AI adoption depends on broader policy conditions, including access to AI‑enabling technologies, data, skilled personnel, and supportive governance arrangements. These elements determine whether border agencies and supply chain actors can deploy AI tools at scale and with confidence.

Realising the full potential of AI and other advanced digital technologies in trade facilitation also requires readiness across the entire supply chain. AI can only deliver efficiency gains when all actors involved – from traders and logistics service providers to customs administrations and other border agencies – operate within a digital environment that supports paperless processes (OECD, 2025[14]). This means that digitalisation must extend across every stage of the transaction, from the creation of commercial documents to the submission of regulatory information and the processing of clearances. Ensuring that the core building blocks for going paperless are in place (Figure 2.1) – digitised documents, streamlined procedures, common data standards, and interoperable information systems – is essential. Only when these components function effectively and cohesively can AI enhance the speed, accuracy, and resilience of border operations.

Figure 2.1. The building blocks for going paperless: A conceptual framework

Source: (OECD, 2025[14]).

The digitisation of trade-related documents, such as invoices, bills of lading or customs declarations is critical. Fragmented and analogue processes prevent AI from being applied to core tasks such as automated document processing or risk assessment. The digitalisation of border procedures is another requirement. For example, digitalised pre-arrival processing1 or digital Single Window systems2 are important pre-conditions for the adoption of AI in trade facilitation systems. A systematic digital procedure that enables domestic and cross-border agency co‑operation can also address the technical interoperability challenges which prevent the actors along the supply chains to exchange risk intelligence and compliance data.

The standardisation of data elements, through the adoption of international standards, such as the UN Trade Data Element Directory (UNTDED/ISO7372), the “buy-ship-pay” data model, the Supply Chain Reference Data Model, and document presentation standards for e-invoicing or e-sanitary and phytosanitary certificates, is essential for establishing interoperability with foreign customs systems and facilitating the development of machine-readable document formats. Finally, the adoption of supportive legal frameworks provides the basis for recognising electronic transactions, governing digital border processes, and allowing trusted data to move across borders. Establishing legal certainty and infrastructure for e-signatures, e-authentication, and secure cross-border data flows is critical in creating digital ecosystem required for AI to function effectively.

However, the deployment of AI in trade facilitation is not without challenges including in the context of AI governance, and particularly regarding trust, accountability, and the ethical use of technology across the supply chain. As emphasised by the OECD AI Principles (OECD, 2024[15]) transparency and explainability enable people to understand how an AI system is developed, how it operates, and how it arrives at specific outcomes, allowing users of AI systems to make informed choices. In the context of paperless trade, a lack of transparency and explainability can, for example, create concerns about how decisions on risk profiling or private-sector logistics optimisation are made.

Maintaining a human in the loop is important to ensure accountability allowing human agents to validate AI generated insights and handle edge cases or contextual nuances that automated systems may misinterpret. The mitigation of bias in data also remains a concern for going paperless with AI. AI models trained on historical trade data risk consolidating existing biases, such as historical bias or selection bias, which can lead to the discriminatory targeting of specific traders, regions, or goods.

2.2.2. The trade policy landscape also needs to be conducive to the adoption and diffusion of AI

The adoption and diffusion of AI for trade facilitation depends critically on the availability and cross-border movement of key inputs, including AI-enabling goods and services, skilled workers, and data. Trade policy, together with the domestic and international policy environment underpinning AI adoption and diffusion, plays a central role in determining whether customs and other border agencies can access and deploy AI at scale.

The AI-enabling trade policy environment has six layers (Figure 2.2 – see (Ferencz, López González and Oliván García, 2022[4])).

First, access to AI goods. Tariffs and non-tariff measures on information and communication technology (ICT) hardware, such as advanced semiconductors and sensors can affect access, increasing the cost of the physical infrastructure underpinning AI.

Second, access to digital services underlie the ability to connect and operate AI systems. Restrictions to critical services, like computer and telecommunications services, can hinder access to digital inputs, including the AI technologies themselves but also cloud computing. They can also raise the costs of connectivity infrastructure on which AI systems operate (Aaronson, 2024[16]).

Third, AI relies heavily on human capital with specialised AI expertise (Goldfarb and Trefler, 2018[17]). Since AI expertise is concentrated, domestic policies affecting the mobility of skilled workers across borders can have a significant impact on a country’s ability to attract the talent needed for developing and adopting AI systems, including at the border.

Fourth, cross‑border data flows. These are the lifeblood of AI. Stringent restrictions on cross‑border data transfers and localisation requirements can hamper the aggregation of diverse datasets required for training, retraining, and fine‑tuning AI models, as well as for enabling real‑time inference.

Fifth, a domestic AI policy framework. These include national AI strategies, legal frameworks, and well‑defined institutional arrangements for AI governance. They provide clear policy direction and greater certainty for AI service providers.

Sixth, international co-operation. This includes increased international discussions on issues relevant to AI, in regional trade agreements and digital economy partnerships, but also in wider fora such as the United Nations, Asia-Pacific Economic Cooperation (APEC)3 or the OECD.

Figure 2.2. Trade policy plays a central role in enabling the adoption and diffusion of artificial intelligence

Source: based on (Ferencz, López González and Oliván García, 2022[4]).

In addition to going paperless and the trade policy conditions that enable access to, and use of, AI tools, the effective deployment of AI at the border also depends on building the capabilities of customs and other border agencies to integrate AI into day-to-day operations. This calls for targeted training and upskilling so staff can interpret AI outputs, understand limitations and potential biases, and exercise informed human oversight. Continuous learning and change management are particularly important to ensure that AI complements existing risk management workflows, strengthens compliance, and supports consistent decision making across agencies.

Notes

← 1.  Procedures that allow for the submission and processing of trade-related documentation and information before the arrival of goods at the border.

← 2.  Platforms allowing traders to submit all necessary documentation and data required for import, export and transit procedures through a single-entry point.

← 3.  The APEC Economic Leaders endorsed the APEC Artificial Intelligence Initiative (2026‑2030) as part of the 2025 APEC Leaders’ Gyeongju Declaration in November 2025, recognising AI’s potential to fundamentally reshape economies worldwide by unlocking new frontiers for innovation, enhanced productivity, improved competitiveness, economic prosperity and resilience.