A Data-Driven Business Models

The automotive industry is in transition where business model changes are concerned.Footnote ²⁵ “Traditional” automotive manufacturers are now transforming into a new mobility ecosystem – from a mass-produced goods-sale business into highly customized data-based service suppliers.Footnote ²⁶ In order to address changing consumer demand, the automotive industry is becoming less and less industrial and, simultaneously, increasingly intent on services, especially the operation and maintenance of vehicles.Footnote ²⁷ Moving toward data-driven business models, AI further shatters the boundaries between the physical and the remaining components of the CAV.Footnote ²⁸

At the same time, another significant transformation in the automotive industry is the use of CAVs to supply Mobility as a Service (MaaS) solutions. Vehicle manufacturers will play an important role in MaaS services.Footnote ²⁹ On the one hand, the transportation environment will be increasingly dominated by car-sharing, ride-hailing, and related services. In this context, being driven by CAVs will represent the essential nature of future transportation, and automotive firms will supply car-sharing and ad hoc use applications to drive up utilization rates. On the other hand, revenues generated by MaaS may become the core source of shareholder value creation when traditional car sales decline.Footnote ³⁰ To summarize, the data-driven economy enabled by CAVs will displace vehicle ownership with MaaS, leading to a new transport system landscape.Footnote ³¹

In light of these changes, business models for CAVs will become increasingly complex. Considering the trends toward MaaS, the typical business model will be based on a function that combines tangible vehicles and intangible services. Automotive firms will offer services to adjust existing functionalities and update software to increase the automated capabilities of a vehicle. Because the economic value of CAVs relies on their use value, throughout their operational life, tangible vehicles and intangible services must be combined to jointly fulfill customers’ needs.Footnote ³² In this respect, it is fair to say that the higher the level of automated driving systems, the more service-oriented the automotive industry. Despite this reality, most CAVs, especially levels 1–4 of driving automation, fall in between traditional “goods” and “services.”

B Integrated Technical Features

CAV systems are highly integrated, both internally and externally. In terms of internal integration, a CAV may contain several driving automation features that have individual, narrow-use specifications, but which together may provide advanced automated driving functions. Technically speaking, a combination of automation features means that an aggregate level of safety is necessary for CAVs. In other words, part-based standards, for example steering wheels, no longer make any sense.Footnote ³³ CAVs should be regulated at the system level to ensure the overall safety of the entire driving system. Only an aggregate “measure of safety” can cope with the challenges of “cumulative” automation features and therefore adequately protect CAV safety.Footnote ³⁴ The “target” of safety regulation, therefore, should be gradually shifted from “auto parts” (goods) to integrated CAVs (overlapping boundaries of goods and services).

In terms of external integration, vehicles are rapidly transforming into “connected devices.” Under the technological trends of V2X, a CAV is merely one component of the entire transport ecosystem. Depending on the level of automation, a CAV may be able to communicate with its occupants, other vehicles, road users, the surrounding transportation physical infrastructure, and all other Internet-based devices and applications. Indeed, in the extreme, in the age of AI, the world we are living in can be described as a convergence of all “IoT devices.” CAVs will always be data-driven – by digitally connecting to one another and their surroundings. By interacting with the external physical and digital environment through V2X communications, CAVs are literally “components” of a holistic transport landscape.Footnote ³⁵ This complex transport ecosystem requires a regulatory framework that considers security convergence, namely the combination of physical security and cybersecurity.

Despite this reality, security risks are particularly complex for CAVs, because they operate across both the physical and the digital world.Footnote ³⁶ Compared to conventional vehicles, risks to a CAV involve threats related to the integrated environment. When communicating with other vehicles and infrastructure, CAVs become a channel for attack and an opportunity for hackers. Hackers can target the CAV itself, the servers supporting it, or the external systems that communicate with the CAV. It is also technologically possible for attackers to seize control of an entire fleet of CAVs by breaching the infrastructure.Footnote ³⁷ In this context, the internal and external integration of CAVs may lead to complex security concerns. There is a need not only for “device” security but also for entire ecosystem security, with a strategic approach to threats. Regulators must ensure that CAVs are safe, both mechanically and in terms of protection from cyber attacks.Footnote ³⁸

C Technical Barriers to Trade Agreement or General Agreement on Trade in Services?

Considering their business models and technical features, CAV-related safety standards may disrupt the scope of coverage of the TBT Agreement. Are CAVs goods or services?Footnote ³⁹ Classification determines whether and to what extent the TBT Agreement rules are applicable. When “goods” and “services” converge as a package in the CAV market, the same is true for relevant safety standards.Footnote ⁴⁰ The integrated CAV system, under which services are embedded with the physical body of the CAVs, is disrupting the traditional boundaries of trade regimes in terms of standards.

It should be noted that WTO case law generally supports the existence of a “boundary” between trade in goods and trade in services.Footnote ⁴¹ The Appellate Body has repeatedly stressed that whether a specific measure is scrutinized under the General Agreement on Tariffs and Trade (GATT), the General Agreement on Trade in Services (GATS), or both is a matter that can only be determined on a case-by-case basis.Footnote ⁴² Here, the Singapore Standard Council, for example, has issued a Technical Reference related to an enhanced cybersecurity framework for CAVs.Footnote ⁴³ The Technical Reference, among other things, requires CAV developers to provide comprehensive documentation for a security-by-design review, to conduct cybersecurity assessment, and to comply with “cybersecurity principles” throughout the full lifecycle of CAVs, including design, development, operations, maintenance, and decommissioning.Footnote ⁴⁴ The purpose of this Technical Reference is to provide rules to govern the cybersecurity assessment framework of CAVs on public roads. Toward that end, drawing from best practices in the industry, the Technical Reference provides standards. These standards apply to “a cyber-physical vehicle system,” which includes embedded control systems and “a coupling between the computational elements and physical elements.”Footnote ⁴⁵ In this particular instance, is the standard under Singapore’s Technical Reference for Autonomous Vehicles a measure of goods or services?Footnote ⁴⁶

Figure 6.1 demonstrates the evolution of CAV standards.Footnote ⁴⁷ At the core of the issue is whether the regulations under the emergence of a new ecosystem of mobility should be subject to the TBT Agreement. Based on the factors delineated earlier regarding the transformation of the auto industry, the MaaS, the cumulative automation features of integrated CAVs, and the complex security concerns involved, the classification or reclassification of CAVs – in the context of the SAE’s six levels (levels 0–5) – will prove an interesting case. At one extreme, level 0 (no automation, human driver), representing conventional, “personally owned” vehicles,Footnote ⁴⁸ should be classified as goods and thus fall within the lower-left quadrant. At the other extreme, level 5 (full automation, “passengers only”),Footnote ⁴⁹ representing shared CAVs, should be classified as services and thus fall within the upper-right quadrant. The remaining levels of CAVs (levels 1–4) comprise both goods and services under this four-quadrant analytical framework. During this time of transitional human–vehicle cooperation, CAV measures govern “trade in goods” but also affect “trade in services.”

Figure 6.1

(Re)classification of connected and autonomous vehicle standards

As shown in Figure 6.1, as CAVs evolve from level 0 to level 5, more and more domestic regulations will be subject to GATS. That said, CAV standards are arguably a “technical standard” within the meaning of Article VI:4.Footnote ⁵⁰ The requirements in GATS Article VI:5, including “reasonable” and “not more burdensome than necessary,” should be further explored in relation to high-level CAV standards.Footnote ⁵¹ However, the GATS contains too few trade rules to handle level 5 CAVs.Footnote ⁵² More importantly, in the interim, regarding levels 1–4, how should CAV standards be reclassified? The obligations under the TBT are far more substantial than those under the GATS. The danger of legal unpredictability may be imminent. To conclude, automation systems are “disruptive” in the way that they challenge existing governance frameworks and disrupt the boundaries of the TBT Agreement.

A Industry-Driven Standardization Process

One noteworthy angle in the ongoing process of CAV standardization is the industry-led approach. Such a regulatory scheme emphasizes market incentives rather than top-down regulation.Footnote ⁵³ Government agencies consider themselves to be in partnership with developers in pursuit of the safe and rapid deployment of CAVs. One outstanding example is the subtitle of the US National Highway Traffic Safety Administration (NHTSA) guidelines: “Accelerating the Next Revolution in Roadway Safety.”Footnote ⁵⁴ By stressing its role as a facilitator in the exchange of information among CAV stakeholders, the US government does not play an exclusively dominant role in shaping the standards of automation systems.

This privatization of governance is attributable, in part, to governments’ lack of requisite technical expertise, as well as the flexibility necessary to address ever more complex regulatory tasks.Footnote ⁵⁵ One of the potential legal hurdles that might slow the deployment of CAVs is bureaucracy, which often moves much more slowly than technological changes. Relevant agencies are aware that CAV development is taking place in a remarkably complex and dynamic technological environment, and that governments are currently in no position to provide mandatory performance standards for these emerging technologies.Footnote ⁵⁶ Indeed, the development of CAV standards requires timely action. The scheme under which the private sector leads the standardization process has proven to be a more effective approach.

Procedurally and practically, the concept of “co-governance” has increasingly been advocated.Footnote ⁵⁷ The CAV ecosystem comprises a variety of interconnected stakeholders, including the automotive industry and software businesses.Footnote ⁵⁸ Governments must govern alongside private and civic sectors in a more inclusive, collaborative, and dynamic manner to drive cross-industry discussion. Through a relatively inclusive and transparent process,Footnote ⁵⁹ governmental agencies participate on an equal footing with stakeholders.Footnote ⁶⁰ In a world in which technological development is firmly in the control of industry, “regulated” parties should be treated as committed partners.Footnote ⁶¹ The collaborative approach may prove perfectly sensible.

B Voluntary Standards

An empirical survey reveals that in most jurisdictions, CAV standards go by many names, including “guidance,” “guidelines,” “recommendations,” “informal standards,” “best practices,” and “codes of conduct.”Footnote ⁶² These informal regulatory mechanisms, which in general are created to be “morally” or “politically” binding, can be considered to fall under the broad umbrella of “soft law.”Footnote ⁶³ This demonstrates how CAVs have been “regulated” in the loosest sense of the term.

Empirical studies also demonstrate that more and more jurisdictions aim to minimize mandatory governmental regulation and favor voluntary, industry-led, nonbinding standards to enhance CAV safety.Footnote ⁶⁴ In the UK, for example, the Automated and Electric Vehicles Act 2018 received Royal Assent on 19 July 2018.Footnote ⁶⁵ The UK’s Law Commission completed its first round of consultation in February 2019, which focused on regulatory frameworks, namely how to ensure safety when using CAVs, and how current road rules should be adapted for AI.Footnote ⁶⁶ The primary considerations identified in the consultation paper identify areas in which there may be ambiguity in the law, as well as potentially necessary reforms. Stakeholders have actively responded to the key questions raised in the consultation paper, including, among other items, how to allocate civil and criminal responsibility when control is shared between the automated driving system and a human user, impacts on other road users, and protection from risks.Footnote ⁶⁷ More importantly, the government has already published a “Code of Practice” to provide “guidance” for CAV testing. Although, by the very nature of the code, compliance is voluntary, it nevertheless sets out “principles” and details “recommendations” that the government believes should be followed to minimize potential risks and maintain safety.Footnote ⁶⁸

On the other side of the Atlantic, the US NHTSA and the US Department of Transportation (DOT) jointly issued the Federal Automated Vehicles Policy in September 2016.Footnote ⁶⁹ The Policy was designed to “set forth a proactive approach to providing safety assurance and facilitating innovation.”Footnote ⁷⁰ The NHTSA issued “Automated Driving Systems: A Vision for Safety 2.0” (Guidance 2.0) in September 2017,Footnote ⁷¹ which, based on the comments of key stakeholders (e.g., the automotive industry) as they considered “best practices,” offers a “flexible, nonregulatory approach” to CAV safety.Footnote ⁷² In October 2018, the DOT released “Preparing for the Future of Transportation: Automated Vehicles 3.0” (Guidance 3.0), which again serves as “voluntary guidance” and “is intended to be flexible.”Footnote ⁷³ As advanced by some commentators, the NHTSA’s CAV guidelines are indeed “an exuberant celebration of volunteerism.”Footnote ⁷⁴

In Asia, one striking case is China’s CAV roadmap. Following the “New Generation Artificial Intelligence Development Plan,” issued by the State Council in 2019, the Ministry of Industry and Information Technology announced a Three-Year Plan for Promoting Development of a New Generation of the Artificial Intelligence Industry (MIIT), implemented to detail action plans for the development of driver assistance systems, vehicle intelligence algorithms, and automotive smart chips.Footnote ⁷⁵ Further, the MIIT has published “communication guidelines,” which cover ninety-two standards, with a focus on CAV technology.Footnote ⁷⁶ The government also announced that more than thirty key standards that are critical to autonomous driving systems will be introduced in the coming years.Footnote ⁷⁷ China’s approach to CAV development is systematic. Following China’s top-down standardization process, the government continues to take primary responsibility in terms of standardization development. However, in the case of CAVs, the government has been working closely with the CAV industry, including the China Industry Innovation Alliance for the Intelligent and Connected Vehicles, to ensure the “relevance and flexibility” of standards.Footnote ⁷⁸ The alliance, together with other industry associations, has been commissioned by the MIIT to develop a common set of “protocols” for CAVs.Footnote ⁷⁹

To summarize, most countries maintain CAV safety policies while emphasizing the voluntary nature of standards and safety assessments.Footnote ⁸⁰ Governments tend to refrain from mandating CAV-specific design features and performance standards. In addition, relevant authorities are inclined to offer nonregulatory approaches to CAV safety. CAV developers may “consider” the guidance as they develop, test, and deploy CAVs on public roadways. They are also “encouraged” to submit a “safety self-assessment,” describing their treatment of each guideline.Footnote ⁸¹ This chapter is not the place to provide a detailed analysis of whether existing approaches can meet CAV safety needs. However, a few general comments bear emphasis. To some extent, the voluntary approach seems realistic. Standards of a soft law nature offer advantages over traditional command-and-control regulation because they provide greater flexibility and adaptability and lower compliance and administrative costs, directly address industry-specific and consumer issues, and adapt to the rapidly changing political landscape.Footnote ⁸²

C Technical Barriers to Trade Agreement: In or Out?

The proliferation of “soft” safety standards in the CAV industry reveals a pattern of self-regulation. In 2018, the SAE formed a committee of stakeholders, took the lead, and published the CAV safety “principles.”Footnote ⁸³ In 2019, industry leaders across the CAV technologies also published “Safety First for Automated Driving,” a nonbinding framework for the development, testing, and validation of safe CAVs.Footnote ⁸⁴ The distinguishing feature of these standards is their legally nonbinding nature, which strongly correlates to the self-regulation of nonstate actors. Indeed, they were literally drawn up by the private entity that is to be regulated.Footnote ⁸⁵

This ongoing shift to voluntary co-governance raises an important question: Under what conditions can the complaining party invoke the dispute settlement system against nonbinding CAV standards that have developed by the private entities?Footnote ⁸⁶ In Figure 6.2, the vertical axis represents the relative level of governmental involvement in the standardization process, while the horizontal axis represents the relative degree of the binding effects of the standards. At the crux of the matter is this: the emerging CAV standards have implications for the boundaries of the TBT Agreement. First, the fact that the CAV standardization process may lack sufficient governmental involvement raises the question of whether the TBT Agreement will apply. Second, these CAV standards may fall outside the definition of “technical regulation” in the TBT Agreement. As clearly argued by Lim in Chapter 5 of this book, the first issue is more important than the second, because although the TBT disciplines on technical regulations are relatively deeper, voluntary standards are still subject to Article 4 and Annex 3 of the Agreement.Footnote ⁸⁷ Nevertheless, as shown in Figure 6.2, the transformative nature of disruptive innovation renders the interpretation and application of TBT rules challenging.

Figure 6.2

Co-governance of connected and autonomous vehicle standards

To illustrate, the WTO provides a dispute settlement mechanism whereby a member considers that its benefits under the covered agreements are being impaired by “measures” taken by another member.Footnote ⁸⁸ In this context, any act or omission “attributable” to a WTO member can serve as a “measure” of that member for purposes of dispute settlement proceedings.Footnote ⁸⁹ “A complainant must clearly establish that the alleged measure is attributable to the responding Member.”Footnote ⁹⁰ The decisive factor here is “the degree of governmental involvement.”Footnote ⁹¹ WTO case law indicates that “private actions” may fall within the scope of the TBT Agreement if the support provided by a government is sufficient to become a “governmental act.”Footnote ⁹² Considering the approaches taken by WTO panels and the Appellate Body in past disputes,Footnote ⁹³ the conduct of the private body may come under WTO disciplines if there is a demonstrable link between the government and such “conduct.”Footnote ⁹⁴ To summarize, a nexus must exist between the responding member and the “measure” – regardless of whether it is an act or an omission.Footnote ⁹⁵

At the core of the question is whether the standards published by a nongovernmental body at the request of the government, or with some degree of government support, can be viewed as a measure of a member government.Footnote ⁹⁶ Is there an adequate connection between a private entity’s self-regulation and government action? To what extent should a tribunal impose the responsibility of WTO members with regard to industry standards? The emergence of the self-regulating, collaborative approach for CAV standard-setting therefore leads to the question of how the TBT Agreement should respond to the trend of public–private convergence in the standardization process.Footnote ⁹⁷ As shown in Figure 6.2, the CAV case study demonstrates that the co-governance model, in which governments “more or less” “work together” with stakeholders of the CAV ecosystem, arguably falls into the middle-lower portion of the “Governmental Involvement” axis.

Furthermore, even if sufficient governmental involvement can be found in the CAV standard-setting process, the next question is whether these “standards” can constitute “technical regulation” within the meaning of the TBT Agreement, therefore allowing a set of deeper trade disciplines to apply.Footnote ⁹⁸ In past WTO cases, the panels and the Appellate Body elaborated on the term “mandatory,” indicating that it includes “binding as well as obligatory, compulsory, not discretionary,” or “required by law or mandate; compulsory.”Footnote ⁹⁹ However, as clarified in several disputes, “the mandatory character of the measure may result from a combined effect of various measures or actions attributable to the Member in question.”Footnote ¹⁰⁰ In other words, a “voluntary and non-mandatory” measure may become “mandatory” as a result of “some other governmental action” or “some other action attributable to the Member concerned.”Footnote ¹⁰¹ As shown in Figure 6.2, the industry-driven, voluntary standards of CAVs may be determined to fall outside the definition of “technical standards” under the TBT Agreement because of its private and soft nature, unless compliance with the standards becomes de facto mandatory because of “some other action attributable to the Member concerned.”

That said, CAV standardization is indeed an interesting case study in the determination of “de facto mandatory.”Footnote ¹⁰² In terms of administrative actions, CAV safety guidance, although industry-led and nonbinding, may become a core requirement for “duty of safety” and may also have important evidentiary value in regulatory investigations. The UK’s CAV consultation paper, for example, indicates that “there is a rebuttable presumption that a product is safe if it conforms to voluntary standards published by the Commission.”Footnote ¹⁰³ In terms of judicial litigation, depending on the level of automation, courts need legal standards to determine liability when CAVs crash. More specifically, for levels 1–4, in which humans and CAVs are codrivers, determining “cause” during the transition to a driverless future will become increasingly complex in lawsuits related to negligence or product liability. Did the CAV properly alert the human driver?Footnote ¹⁰⁴ Should the CAV have been designed to automatically reduce vehicle speed on a snowy road? Or should the system prevent driving altogether?Footnote ¹⁰⁵ In a negligence or product liability lawsuit involving CAVs, the key issues will be whether the design of the CAV is defective in nature. This will inevitably lead to an inquiry into the “standard of care” that is imposed on CAVs, or the definition of a “design defect” for CAVs.Footnote ¹⁰⁶ How do manufacturers address “reasonable care” when designing the automated driving system? What safety standards should apply to CAVs?Footnote ¹⁰⁷

In practice, the industry-led voluntary standards provide a baseline for judges in the evaluation of appropriate levels and evidence of CAV safety prior to deployment,Footnote ¹⁰⁸ which may become a strong incentive for CAV manufacturers to comply with “guidance,” “best practice,” or “codes of conduct,” thereby demonstrating a commitment to meeting expert safety expectations.Footnote ¹⁰⁹ More specifically, compliance with industry standards can provide convincing evidence regarding whether there is a defect.Footnote ¹¹⁰ CAV manufacturers may be able to partially mitigate the legal risk by demonstrating conformity to industry safety standards. At the same time, judges may resort to industry standards when evaluating the duty of safety in tort cases.Footnote ¹¹¹ In brief, self-regulation is being afforded legal status through complementary evidence.Footnote ¹¹² As a result, private soft law is no longer merely a self-imposed corporate obligation. It can also constitute a source of law in court proceedings.

To conclude, compliance with CAV standards may become de facto mandatory, which somehow blurs the line between mandatary/voluntary standards. The key question, however, remains: Is the TBT Agreement capable of addressing de facto mandatory “self-regulation”? To what extent should a WTO tribunal assume the responsibility of members with regard to CAV safety standards that are prepared and published by a private entity? How should the TBT Agreement respond to the trend of private standardization processes in which the government involvement per se is often minimum, if not nominal? Evidently, the development of disruptive innovation inherently involves changes in governance frameworks and calls for new governance approaches that break the boundaries of existing trade disciplines.