Autonomous vehicles will be an essential part of the mobility of the future. Cars can already relieve the driver in many situations, and the R&D sector for autonomous vehicles is booming. Companies are investing in sensor and machine-learning technology, creating pilot programmes to test self-driving vehicles at levels 4 and 5 of automation. But the export of some of these technologies may be restricted due to potential military applications.
Vehicle automation is divided into several levels. The first three levels imply a non-autonomous vehicle, as the driver is responsible for monitoring the environment. At higher levels (3, 4 and 5) the environment is monitored by a computerised on-board system, supported by various cameras and sensors. From automation level 4 onwards, the driver can take control of the vehicle, but does not necessarily have to, as the car controls all aspects of driving the vehicle.
It will probably be a few more years until we can drive fully autonomous cars. However, the level of vehicle automation and access to increasingly sophisticated driver assistance features is steadily growing. Self-parking, avoiding traffic jams, lane-keeping, blind-spot monitoring, speed limitation systems: we owe all of these automated functions in new cars to modern software, powerful computers, and numerous environmental perception systems such as cameras, radars and laser scanners (Lidar). Now, one of the key technological challenges is to perfect deep-learning algorithms to allow collision-free navigation along a designated route.
However, export controls have not kept pace with technological progress. First, states must establish which products are subject to an export ban or licence, and second, effectively supervise the operation of the control system.
It is important to remember that transfer of technology and know-how doesn’t mean just physical transfer of products, but mainly involves intangible transfer of technology (ITT), posing a serious challenge to any control effort. ITT encompasses transfer by electronic means (e.g. sending an email, providing information over the phone or during a virtual meeting), as well as the transfer of knowledge and skills (e.g. technical assistance, research papers presented at scientific conferences, etc).
In the European Union, the export control regime for dual-use items (i.e. items that can have both military and civilian uses) is based on the Dual-Use Regulation (Council Regulation (EC) No. 428/2009 of 5 May 2009 setting up a Community regime for the control of exports, transfer, brokering and transit of dual-use items). The list of dual-use items forming an integral part of the regulation contains several categories of export control numbers (ECNs) relevant for autonomous vehicle technologies, including among other things semiconductor devices and integrated circuits (ECN Category 3), encryption devices and software (ECN Category 5), cameras and sensors (ECN Category 6), and GPS devices and radars (ECN Category 7).
Depending on its parameters, the export of a particular product, technology or related service may require a licence.
However, many new technologies and services related to autonomous vehicles are not included in the list of dual-use items or are subject to non-binding interpretations. Therefore, exports of technologies related to research and development in the field of autonomous driving rarely require obtaining export licences.
But this situation may change, and the first steps in this direction have already been taken in both the United States and, to a lesser extent, the EU. And the US regulations are already exerting an impact on the export obligations of European entities.
Changes in export controls in the US for emerging and foundational technologies
The changes implemented in US policy over the past few years could significantly hinder the export of autonomous vehicle technology from the United States. In light of the growing importance of certain new technologies to security and defence, and concerns about China’s influence in this area, in 2018 the US introduced the Export Control Reform Act (ECRA).
ECRA is intended to restrict the export of “emerging” and “foundational” technologies with the potential of being used for civilian and military purposes, which have not historically been subject to export controls.
Targeting emerging technologies, the first group of control measures went into effect in January 2020. They imposed licensing requirements on export and re-export to all countries (except Canada) of US-originated software specifically designed for AI-powered geospatial imaging, used, among other things, in autonomous vehicles. In January 2021, it was decided to extend this temporary control measure for another year. Also, discussions are underway to reach an agreement with other countries to impose parallel control obligations.
Persons operating or having access to US-originated software used to train AI systems for image recognition should review the technology they possess to determine whether it is covered by the new classification, and consider the potential licensing lead time when planning procurement schedules.
These regulatory changes may have a direct impact on the obligations of exporters involved in the development of autonomous driving, and the introduced measures have the potential to lead to fragmentation of global supply chains and research and innovation networks.
Indeed, it is important to remember that the US rules governing trade in dual-use products (Export Administration Regulations, EAR) have a broad scope of application. They regulate not only exports from the territory of the United States, but also the movement of US-origin goods and technology between third countries or between nationals of other countries.
Products are subject to US export-control regulations if they are goods, software, or technology located in the US, manufactured in the US, or admittedly outside the US, but the content of US-origin components or technology exceeds the de minimis level specified in the regulations (25% in most cases). A similar regulation applies to certain products consisting of a direct product of US-origin technology (i.e. products developed directly using US-origin technology or software).
EU is modernising the export controls of dual-use goods
The European Union has been working on new regulations on export of dual-use goods for several years, although their scope and impacts will not affect the export of automated vehicle technology as much as regulations introduced by the US.
In November 2020, the European Parliament and the Council of the EU reached agreement on a draft amendment to Regulation 428/2009. The new regulation is to strengthen the EU’s export control toolbox so that the EU can address the risk of human rights abuses associated with trade in cyber-surveillance technologies, as well as gain greater control over trade flows in new and emerging technologies of particular importance.
The very definition of dual-use goods will change to reflect the emergence of cyber-surveillance technology. The definition of an exporter will also change to include individuals and researchers involved in the transfer of dual-use technology.
Stricter export control rules will be introduced for specific cyber-surveillance technologies whose misuse may lead to serious human rights violations and security risks. On the other hand, new EU General Export Authorisations for dual-use products (EU GEAs) are planned for intra-group transfers of software and technology (007) and encryption items (008).
The revised draft of the EU regulation still has to be adopted by the European Parliament, but its entry into force is planned for the first half of 2021.
Autonomous vehicles can bring about many economic and social changes in personal and commercial transport, public infrastructure, urban planning, and supply chain management. It appears that if work on the implementation and diffusion of autonomous vehicle technology does not slow down, the scope of export controls associated with encryption, machine-learning, and artificial-intelligence technology will increase proportionately.
Automotive suppliers, original equipment manufacturers, and other entities involved in the development of automated and autonomous vehicles, among other things, through creation of machine-learning algorithms, should systematically review and update their security policies to ensure they comply with export control regulations.
Anna Olejniczak-Michalska, attorney-at-law, Reprivatisation practice, Private Client practice, Wardyński & Partners