1 Big Data and Cloud Computing

Over the past few decades, an immense amount of data has been generated through the (cross-border) flow of information, humans, products, services, and capital. These developments have resulted in data protection concerns, as well as the specific problem of users of online goods and services “paying” for products offered by disclosing data without assessing the nature and value of the data. The recent European Union (EU) Directive on digital content even regulates the payment of online services through the provision of data.Footnote ¹⁶ Furthermore, data holders are often unaware of how much data is collected and stored about them.

Big data is a term coined for datasets whose size is beyond the ability of commonly used software tools to capture, curate, manage, and process within a tolerable degree of time. The phenomenon of big data analytics is often characterized by four elements, namely volume, variety, velocity, and veracity (the “4 V”). “Volume” refers to the especially large amount of data; “variety” makes it clear that the data is derived from manifold sources and formats; “velocity” mirrors the high speed of the data processing; and “veracity” reflects the reliance on the correctness of the data. A special feature of big data from a legal perspective is the fact that the traditional concept of causation is replaced by the concept of correlation.Footnote ¹⁷

For some years, international organizations such as the Organisation for Economic Co-operation and Development (OECD), as well as academic voices, have assumed that big data is potentially a key driver of innovation, productivity growth, and economic competitiveness.Footnote ¹⁸ The global exchange of data requires unrestricted cross-border data flows in order to realize its merits. In this context, big data analytics is able to improve the outcome of data processing in manifold areas.Footnote ¹⁹ Examples include the automotive industry, which evaluates ideas submitted through its “virtual innovation agency,” and the health sector, which collects data on larger populations in order to reduce disease, bringing rapidly and accurately identified drugs to the market and providing better healthcare by enabling the application of evidence-based interventions.Footnote ²⁰

A further technical innovation is cloud computing, which facilitates cross-border data flows in order to take advantage of cheaper on-demand computer capacity that can be scaled and paid for as needed. Cloud services encompass software and access to processing, email, storage, and other computer resources. From a business perspective, cloud computing turns a fixed IT cost into a variable operating cost, and from the angle of human resources, cloud computing underpins the capacity of people to work remotely.Footnote ²¹ Major challenges include connectivity and the compliance of cross-border data flows with data protection and data security requirements.Footnote ²²

2 Artificial Intelligence

Algorithms are able to automate the “production” of goods/services and to facilitate the selection and filtering of information in various ways, thereby attributing relevance to certain data and moderating content. Algorithms are the foundation of AI as a machine-based system that is capable of influencing the environment by making recommendations and predictions without human input.Footnote ²³ An AI system lifecycle encompasses various phases, such as the design and modeling of data, the verification and validation of data, as well as the deployment, operation, monitoring and storage of data.Footnote ²⁴

AI allows for the implementation of a “regime” of automated decision-making to be conducted in a highly timely and effective manner. Such automation is primarily feasible in situations that do not require a specific human intervention. The new technologies can also transform traditional manufacturing into smart manufacturing, focusing on digitization from early product design through maintenance at the product’s end of life by using advanced sensors and big data analytics.Footnote ²⁵ As examples, and only at the very beginning of data-driven innovation potential, the following phenomena can be identified: (i) driverless cars will become a reality, and (ii) robotics will move forward to become a widespread industry tool based on rapidly advancing AI.

Notwithstanding the fact that AI has many benefits, some risks also cannot be overlooked. Therefore, from a normative concept of civil society, a few questions must be considered in an interdisciplinary manner:Footnote ²⁶ (i) Do AI processes comply with fundamental constitutional principles? (ii) Is the AI application based on a sufficient legal foundation? (iii) Do AI processes comply with the applicable requirements of data protection laws? (iv) Who is in charge of monitoring the socially responsible use of AI, and who is liable in case of a failure caused by an algorithm?

These questions merit appropriate answers, even if solid responses that serve the needs of global society are not easy to find. Nevertheless, a reconciliation of AI’s chances and risks must be kept in mind if an adequate legal framework for digital governance in global relations, particularly in the international trade regime, is to become a reality.

1 Globalization and Governance

Globalization is not a clearly defined term; its concrete meaning depends on the given substantive component of societal life.Footnote ²⁷ (i) Legal globalization concerns the harmonization of states’ normative orders, or the implementation of cross-border legal rules. (ii) Cultural globalization addresses those issues related to manifold social policies. (iii) Commercial globalization reflects the existence of increased transnational businesses and economic activities. Global law as an aspect of legal globalization is confronted with new concepts, examining institutional differentiations and elaborated procedural techniques.Footnote ²⁸

The term “governance,” which stems from the Greek word “kybernetes” and the Latin word “gubernator,” means a steersman, and it must be recognized for its importance. Governance can be addressed from the perspective of different disciplines; nevertheless, at whatever level of social organization it may take place, governance refers to the appropriate business conduct of a private or a public body. In this context, some key questions must be asked and answered:Footnote ²⁹ (i) Who is entitled to set the rules? (ii) In whose interest are the rules? (iii) By which mechanisms are the rules created? (iv) For which purposes are the rules designed? There is a need to develop overarching networks and negotiation systems between different stakeholders, thus forming a “cooperative approach to governance” that includes the entire society, hence dividing responsibilities between public and private actors.

Based on the described notions of “globalization” and “governance,” the development of a broader rule-making approach encompassing the needs of the international trade regime and of the digital environment of today’s societies appears to be unavoidable and equally justified.

2 A Broader and Better-Coordinated Rule-Making Approach

Assessing the dichotomy of regulatory sources and the emergence of new regimes introduced by civil society, adapted transnational concepts must be developed in the global law environment.Footnote ³⁰ Institutions can lead states to more cooperative behavior than they otherwise might have adopted, building mutual connections from peripheral points in federative or associate forms.Footnote ³¹ In realizing an appropriate governance framework, civil society involvement should be strengthened. The facilitation of networking opportunities and public support from concerned persons and organizations in the relevant policy field should also be considered; that is, a policy vision must be developed.Footnote ³²

In the Internet governance context, a new approach has been developed and applied in part, namely the multistakeholder participation model, which attempts to involve all concerned persons and organizations in the public and private sphere in the discussions and negotiations of the regulatory framework.Footnote ³³ Practical experience has shown that some basic challenges need to be addressed in order to ensure that the multistakeholder concepts are successful. Four fundamental questions must be tackled:Footnote ³⁴ (i) How do governance groups best match challenges with organizations and networks? (ii) How can governing bodies/entities be most able to help develop legitimate, effective, and efficient solutions? (iii) How should the flow of information and knowledge necessary for successful governance be structured? (iv) How can different governance groups approach coordination between available governance networks in order to avoid conflicting interests?

Over the last few years, the globalization of the international legal framework, among other fields in international trade law (particularly due to the “outdated” classification regime in the World Trade Organization (WTO) Agreements), and equally in the area of Internet law (due to a stronger emphasis on state sovereignty), has suffered some setbacks. In light of the fact that the legal fragmentation caused by national laws jeopardizes an appropriate design of global law in a network society, coherence between different initiatives should be strengthened in order to overcome trends leading to various forms of fragmentation or distorted regulatory regimes.

The recently developed term “legal interoperability” addresses the process of creating legal rules that cooperate across jurisdictions.Footnote ³⁵ This objective can be realized in a matter of degrees, as many options exist between a full harmonization of normative rules and a complete fragmentation of legal systems.Footnote ³⁶ As is so often the case in the real world, striking the correct balance is of utmost importance. While an excessively high level of interoperability could cause difficulties in the management of the harmonized rules and fail to acknowledge social and cultural differences, a level too low could present challenges to smooth social interaction.Footnote ³⁷

1 Transparency

All involved stakeholders should promote a culture of transparency, enshrining the disclosure of data logics and access to the applied algorithms and datasets.Footnote ³⁸ Transparency is usually defined as “easily seen through … evident, obvious, clear.”Footnote ³⁹ Transparency means understandable and forward-looking information, appropriate to the context and the state of the art, in order to make stakeholders aware of their interactions (in an ex ante or ex post data-centered decision-making process).Footnote ⁴⁰

Transparency requires robust and general rules, not necessarily more regulation; that is, the improvement of transparency does not mean a quantitative increase in information, but rather “more” in terms of higher information quality.Footnote ⁴¹ A future-oriented understanding of transparency should observe the following elements:Footnote ⁴² (a) the existence of publicly reliable information; that is, substantive quality standards related to information, supported by an adequate legal framework; (b) the designation of the information recipient as a holder of rights and an essential component for perception and transparency; and (c) the availability of disclosure procedures and observance of the time element; that is, transparency implies constant visibility of information.

Providing information about the type of input data and the expected output, explaining the variables and their weight, and shedding light on analytics architecture usually contribute to transparency with respect to AI algorithms.Footnote ⁴³ Nevertheless, a generic statement on the use of AI does not allow for the easy assessment of all challenges and risks; concrete circumstances do play a role, which means that solutions focused on disclosing specific information about the applied algorithms may be the best option.Footnote ⁴⁴

2 Accountability

All stakeholders involved in datafication and AI mechanisms should be accountable for the proper functioning of the systems employed, as well as the integrity of the regulatory environment.Footnote ⁴⁵ Accountability helps to ensure an environment in which individuals and enterprises assume their respective responsibilities. The first legislative attempts to meet this requirement can already be seen: for example, in the field of data protection, the EU General Data Protection Regulation calls upon organizations to apply a “Privacy by Design/Default” approach and – under certain circumstances – to conduct a “Data Protection Impact Assessment.”Footnote ⁴⁶

Accountability encompasses the obligation of one person to another, according to which the former must give an account of, explain, and justify his/her actions and decisions against criteria of the same kind.Footnote ⁴⁷ Therefore, the proportionality principle, which inspires an adequate and appropriate deployment of big data analytics and AI, should apply.Footnote ⁴⁸ Accountability also relates to good governance, which was previously addressed. The development of respective concepts in public institutions and private enterprises requires publicly assessable accounts as a precondition for a sustainable society.Footnote ⁴⁹

The obligation to be accountable encompasses the task of disclosing information about the actual AI processes. To improve respective foreseeability, standards should be developed and introduced that moderate the behavioral requirements in a concise manner. Furthermore, the responsibility of the accountable person to ensure that concerned individuals are protected from damages when having suffered a detriment is to be legally developed in a precise way.Footnote ⁵⁰

3 Safety and Robustness

Technological innovations must be safe and robust throughout their entire lifecycle so that data-driven communications and transactions can overcome adverse conditions or foreseeable potential misuse. Safety and robustness also extend to the terms of security and resilience. Therefore, the traceability of the datasets, processes, and decisions must be secured. Furthermore, the risks with respect to safety and robustness should be managed throughout the entire process of hardware or software applications. Consequently, the execution of impact assessments with respect to technological risks is necessary.Footnote ⁵¹

International instruments already state that the likely impact of AI on civil society must be adequately taken into account in order to safeguard fundamental rights. For example, the Recommendation of the OECD Council on Artificial Intelligence refers in part 1.4 to robustness, security, and safety as follows:Footnote ⁵²

Furthermore, the use of AI through modern data-processing techniques and the trend toward implementation of data-intensive processes require a more advanced understanding of risk assessment by individuals, businesses, and public organizations, since possible adverse outcomes stemming from data processes cannot be excluded.Footnote ⁵³ Risks and compliance assessments are not only justified by collective social and ethical values, as well as the nature of fundamental rights and treatments affected by AI application. They also represent an opportunity to better foster public trust as a key objective of the information society.Footnote ⁵⁴

4 Interim Assessment

As stated earlier, the use of modern data-processing techniques and the trend toward implementation of information-intensive data analytics require a more advanced understanding of risk assessment. In particular, the fact that automated decision-making may have an impact on fundamental rights (including the nondiscrimination principle), as well as collective social and ethical values, must be addressed.Footnote ⁵⁵ AI programs can affect various human rights, i.e. the right to human dignity, the principle of nondiscrimination, privacy rights, and the guarantee of self-determination.Footnote ⁵⁶

Furthermore, risks and compliance assessments are not only justified by the nature of the rights and freedoms potentially affected by datafication (or big data analytics), as well as AI applications. In a participatory environment such assessment can contribute to an increased level of trust.Footnote ⁵⁷ As the most recent political developments in the political arena have shown, trust plays an important role in the context of the international trade regime. Trust can even be seen as a central pillar of the globalized governance, as previously discussed.

1 Anticompetitive Behavior

From the perspective of competition law, different issues are at stake. An initial aspect concerns the changed market parameters. Digital markets, as well as the exchange of communications in a digital society, should increase the possibilities for participation among all interested individuals and organizations/businesses. However, the markets tend to be dominated by a few firms (Alphabet (Google), Amazon, Facebook, and Apple (GAFA)). Similar experiences are found in East Asia (Alibaba, Tencent).

The overwhelming dominance of a few players in digital markets causes anticompetitive concerns, which are primarily due to the risk that a market-dominant position has been misused.Footnote ⁶¹ The oligopolistic market structure is now challenged by competition authorities (for example, the Directory Competition of the EU), as well as political bodies in the United States and Europe. The outcome of these interventions is unresolved for the time being.

Other behavioral problems are also on the horizon – for example, the tacit collusion by big firms through the use of parallel algorithms.Footnote ⁶² For good reasons, primarily the OECD, as one of the most important international organizations in the economic field, is thoroughly analyzing the respective challenges.Footnote ⁶³ So far, competition law does not appear to be fully fit to tackle these problems. The lack of general competition law principles in the WTO framework must be seen as another disadvantage for the international trade regime.Footnote ⁶⁴ Furthermore, the lack of coherent competition policies among jurisdictions leads to the disadvantage that national competition authorities are ill equipped to effectively address the anticompetitive data practices; the need for more streamlined standards between antitrust regimes is obvious.

2 Denial of Network Neutrality

A second issue is network neutrality: all market participants (providers of goods and services, as well as consumers) should have unfettered access to the digital infrastructure.Footnote ⁶⁵ The preferential treatment of some businesses or individuals results in the risk of competition distortion. Therefore, discrimination toward certain market participants must be considered unjustified.

In some countries (for example, in the EU), network neutrality provisions do exist.Footnote ⁶⁶ The national implementation of the neutrality principle, however, does not always cover existing needs. In the United States, the trend has moved away from regulatory intervention because of the existing political climate. Obviously, rules cannot replace the market. However, illegitimate discrimination does have a substantial negative effect on the international trade framework.Footnote ⁶⁷

3 Data Localization

A third challenge, more due to efforts of governments than private enterprises, concerns data localization requirements. Around the globe, such requirements are implemented in a variety of forms.Footnote ⁶⁸ Information and communications technology companies may be obligated to host all subscriber and consumer data locally within the country; in some instances, only information covering certain substantive areas (for example, health) must be stored in the country.

Data localization reduces access to data and digital technologies and may also be counterproductive. Requiring data localization in relation to cybersecurity increases data vulnerability in a single jurisdiction, making it easier to target and possibly preventing data backups in globally distributed data centers.Footnote ⁶⁹ Obviously, such provisions raise the costs of access to, and use of, data, thereby reducing gains from digital trade.

4 Interim Assessment

As the aforementioned deliberations demonstrate, global law is exposed to “data demagogues” who strive to interfere with cross-border data flows. Various forms of unjustified market interventions, such as anticompetitive behavior, denial of network neutrality, and data localization provisions, can have a competition distortion effect. Such an outcome must be avoided or, at the very least, mitigated by the regulator.

In addition, legal uncertainty, caused by technological advances, leads to discretionary power that is not always consistent with the rule of law.Footnote ⁷⁰ This fact primarily concerns governments and other regulatory bodies. However, private enterprises can also have power in factual forms. A balance between differing interests is often difficult to find, but efforts to reconcile such interests appear to be unavoidable.Footnote ⁷¹

1 Outdated Goods and Service Classifications

So far, a commonly agreed-upon definition of what represents digital trade does not exist. The WTO has addressed the “production, distribution, marketing, sale, or delivery of goods and services by electronic means” as part of digital trade in its Work Programme on Electronic Commerce of 1998.Footnote ⁷² Cross-border data flows enabling digital trade have not played a major role in the previous discussions, notwithstanding the fact that their appearance is important.

Legal scholars of WTO law have been tackling the distinction between goods (General Agreement on Tariffs and Trade, GATT) and services (General Agreement on Trade in Services, GATS) for many years, since it is not clear which “product category” and consequently which Agreement would be suitable in addressing digital assets.Footnote ⁷³ The differentiation is practically important, since the GATT offers less room for maneuvering to member states that are unwilling to liberalize digital markets than the GATS.Footnote ⁷⁴

The WTO itself is well aware of the existing problems, and efforts have been initiated to remedy the situation. In the context of the 11th Ministerial Conference (December 2017, Buenos Aires), a short “Joint Statement on Electronic Commerce” was adopted, inviting participating member states to undertake further work on digital trade.Footnote ⁷⁵ Most developed countries have agreed to this Joint Statement; China did it almost at the last minute. However, no concrete outcome can be seen for the time being, even though several proposals have been submitted and discussed in the WTO forum. The World Trade Report 2018 of the WTO also intensively addressed digital trade, pointing to the further transformation of the global trade regime. As a result of these investigations, the WTO has pleaded for a technology-induced reshaping of the regulatory environment.Footnote ⁷⁶ Nevertheless, a solution to the existing problems may not surface in the near future.

2 New Data-Oriented Regulatory Approaches in World Trade Organization Law

An important component of the regulatory challenges concerns cross-border data flows. Foundational principles of data regulation in international trade law must be developed.Footnote ⁷⁷ The elements encompass fostering digital trust and ensuring interoperability and transparency, in support of the free flow of information. Therefore, a hybrid regulatory approach based on a polycentric governance modelFootnote ⁷⁸ appears to be suitable.

New business models must become part of the trade rule framework. For example, digital platforms should be able to overcome barriers that have prevented small companies from participating in international trade, and from facilitating the building of trust in global transactions.Footnote ⁷⁹ In summary, regulatory support for the implementation of digital platforms that are stable and trustworthy is an objective that must be pursued.

The WTO rules typically do not refer to private standards, industry best practices, or multistakeholder institutions.Footnote ⁸⁰ However, these models are commonplace in the digital world and therefore must also be incorporated in international trade law. Self-regulation and – possibly – coregulation do have important merits; therefore, it appears to be worthwhile for WTO bodies to cooperate with international organizations and associations that are involved in the development of international trade standards.

Another issue concerns ongoing discussions about digital trade barriers.Footnote ⁸¹ The respective objectives are often at least partly founded in sound policy or social reasoning. Examples include cross-border cooperation on cybersecurity issues, the implementation of an appropriate privacy framework, and the integration of consumer trust-enhancing measures.Footnote ⁸²

Finally, legal scholars are generally of the opinion that the WTO framework should be expanded in the light of smart goods and smart services, constituting the Internet of Things.Footnote ⁸³ Apart from horizontal obligations on cross-border data flows and data localization,Footnote ⁸⁴ efforts must be undertaken to amend the international trade regime by way of new provisions that enable digital innovation and promote business trust.Footnote ⁸⁵ For example, technical standards for digital services that are consistent with internationally recognized standards should be adopted.