Introduction

Artificial intelligence (AI) is poised to reshape international trade by reducing traditional barriers, creating new forms of cross-border exchange, and challenging existing trade rules. AI is already lowering trade costs through applications such as data analytics, translation services, and logistics optimization. At the same time, the development and deployment of AI raise new governance challenges – especially around access to data, privacy, standards, and intellectual property – that international trade rules are increasingly called upon to address.

AI will influence trade not only by changing what is traded and how it is traded, but also by shaping productivity growth, global value chains, and the balance between goods and services in the global economy. Trade agreements, in turn, can either enable or constrain AI development depending on how they handle data flows, regulatory cooperation, and technology protection.

What do we mean by artificial intelligence?

Narrow AI and general AI have a distinction critical for understanding their trade impacts. Narrow AI refers to task-specific applications such as machine translation, chatbots, image recognition, and autonomous vehicles. These systems are already commercially deployed and are driving current trade effects. By contrast, general AI – defined as “self-learning systems that can learn from experience with human-like breadth and surpass human performance on all tasks” – remains speculative and raises broader existential concerns rather than near-term trade issues.

Narrow AI is primarily built on machine learning, which uses large volumes of data and algorithms to generate predictions. This includes supervised learning (using labeled data), unsupervised learning (detecting patterns in raw data), and reinforcement learning, where systems actively generate their own training data. A key technological breakthrough enabling AI’s economic relevance is the Deep Neural Network (DNN), which allows multiple layers of data processing to be combined across text, images, audio, and video. This has produced general-purpose machine learning, enabling AI to function across diverse and complex data environments.

Crucially, data quantity and diversity matter. Access to large datasets – including the ‘tails of data,’ or rare and irregular observations – is essential for training robust AI systems. This dependency on data underpins many of the trade policy issues discussed here.

The impact of AI on economic growth and international trade

AI’s most fundamental trade impact may come through its effect on productivity and economic growth. If AI boosts productivity, it can raise incomes and expand opportunities for international trade. Global productivity growth has been persistently low, in part because economies take time to absorb and reorganize around complex general-purpose technologies like AI. Complementary investments – in skills, capital, and organizational change – are necessary before AI’s benefits are fully realized.

AI is also expected to reshape the composition of growth. AI is likely to accelerate the shift toward services-based economies, with direct implications for trade. Automation threatens low-skill manufacturing jobs, while increasing demand for high-skill services such as data analytics, software, design, and logistics. As AI adds value to products through embedded services, the services share of global trade is likely to expand further.

This transition could widen disparities between countries depending on their capacity to adopt AI, develop skills, and integrate into digitally enabled trade networks.

Specific AI applications to international trade

AI and global value chains

AI is already transforming the operation and structure of global value chains (GVCs). Firms use AI to forecast demand, manage inventories, optimize warehouse operations, and improve just-in-time manufacturing. Robotics and AI-driven inspection systems increase efficiency in packing, quality control, and maintenance.

Industry 4.0, particularly the German-led vision of smart manufacturing that integrates sensors, the internet of things, and cyber-physical systems. These technologies allow predictive maintenance, seamless communication across firms, and highly customized production even for small batches. Such developments could deepen and extend GVCs by enabling specialized service providers to participate in discrete segments such as research and development (R&D), robotics, and data analytics.

At the same time, AI could encourage reshoring. Automation and technologies like 3D printing reduce reliance on low-cost labor and long supply chains. This may intensify what Dani Rodrik calls “premature industrialization,” making it harder for developing countries to industrialize through traditional manufacturing-led export strategies.

Trade using digital platforms

Digital platforms represent one of the most visible trade effects of AI. Platforms like eBay have dramatically expanded small business participation in global trade. In the United States, 97% of small businesses on eBay export, compared to only 4% of offline small businesses.

AI-powered translation services amplify this effect. eBay’s machine translation increased exports to Spanish-speaking Latin America by 17.5%, with export values rising 13.1%. To put this in perspective: a 10% reduction in geographic distance is associated with a 3.5% increase in trade revenue, meaning that AI translation effectively reduced trade ‘distance’ by more than one-third.

Trade negotiations

AI also has potential applications in trade policymaking itself. Governments could use AI to model economic outcomes under different trade scenarios, analyze multi-country negotiation dynamics, and anticipate third-country responses. Brazil has already launched an Intelligent Tech & Trade Initiative that uses AI to support trade negotiations, illustrating how AI could become a strategic tool in trade diplomacy.

Developing trade rules to support AI

The importance of data for AI

Access to data is the lifeblood of AI development, making cross-border data flows a central trade issue. Agreements such as the Comprehensive and Progressive Agreement for Trans-Pacific Partnership (CPTPP) and the US-Mexico-Canada Agreement (USMCA) include commitments to allow data to flow freely across borders, supporting AI training and deployment.

Data localization measures, which restrict where data can be stored or processed, reduce the ability to develop AI systems tailored to different markets. Smaller and developing countries are particularly vulnerable, as they lack large domestic datasets and depend on global data access, especially in areas like healthcare AI.

Privacy and AI

Privacy regulations pose a complex challenge. Laws such as the EU’s General Data Protection Regulation (GDPR) restrict cross-border data transfers and limit how personal data can be re-used, potentially constraining AI development. Requirements to minimize data collection and storage directly conflict with the need for large training datasets.

Yet, we argue that there is no inherent trade-off between developing AI and privacy. Trust is essential for data sharing, and strong privacy protections can enable, rather than hinder, AI adoption if designed flexibly. Trade rules could help by promoting mutual recognition of privacy regimes and shared international principles.

Standards and AI

Divergent national standards, particularly for emerging technologies like autonomous vehicles, risk fragmenting markets and raising costs. Commitments in agreements like the USMCA to base domestic standards on international norms can reduce barriers, promote interoperability, and facilitate AI diffusion.

Protection of source code

Requiring access to source code as a condition of market entry is identified as a trade barrier that undermines AI development. Both the CPTPP and USMCA prohibit such requirements, reflecting concerns about forced technology transfer and the need to protect algorithmic innovation.

Intellectual property protection and AI

AI raises unresolved intellectual property questions, especially regarding the legality of copying data to train algorithms. In the United States, fair use doctrines may permit such activity, but many countries lack similar copyright flexibilities. This legal divergence could become a barrier to global AI deployment. Trade agreements have so far been cautious in addressing these issues, leaving significant uncertainty.

AI and trade in goods

Finally, AI development depends on access to physical inputs such as central processing units (CPUs) and graphics processing unit (GPUs), which are essential for training deep neural networks. This underscores the continued importance of reducing tariffs and barriers on high-tech goods as part of an AI-supportive trade agenda.

Conclusion

AI will reshape international trade patterns, institutions, and rules. While AI offers powerful tools to lower trade costs and expand participation, especially for small businesses, it also exposes gaps in existing trade governance. Data flows, privacy, standards, and intellectual property will increasingly define whether AI becomes a force for inclusive global growth or a source of fragmentation. Trade policy will play a decisive role in determining which path prevails.

Access the full white paper