OECD Business and Finance Outlook 2021

6. Managing access to AI advances to safeguard countries’ essential security interests

Abstract

Artificial Intelligence (AI) has potential to resolve many challenges that our societies face, but as with all innovations, foreign acquisitions of some AI applications may raise security concerns.

International investment in established companies is an important vector for diffusion of AI-related technologies across borders. Concerns about implications for essential security interests have led to tighter government control over such acquisitions, with AI-related technologies often explicitly included in the scope of investment screening mechanisms.

Financing of research abroad is a parallel legal avenue to acquire know-how that is unavailable domestically. It can substitute for acquisitions of established companies. Governments have now begun to set out policies to control such transfers, specifically for AI-related areas. As for foreign investment in equity, such policies need to be carefully devised to avoid forgoing the benefits of international research cooperation. Policy principles agreed at the OECD to strike this balance could offer inspiration.

6.1. Managing risk without stifling opportunities: new challenges require new solutions

While the full scope of future applications of Artificial Intelligence (AI) remains unknown, it is certain that these applications will bring transformational change to all aspects of societies. As with many technological innovations, defence and security applications are likely to be early adopters of AI. To avoid that adversaries or hostile states obtain the technology and the associated military edge, governments seek to establish mechanisms that allow them to manage the proliferation of AI applications and the underlying know-how.

International investment is among the most important channels for technology diffusion. To safeguard their essential security interests, governments have consequently placed great emphasis on managing AI diffusion through international investment. Investment screening mechanisms are the main instrument employed for this purpose: They allow governments to review foreign investment proposals in sensitive sectors for potential threats to essential security interests and to impose conditions, or, as a last resort, to block or unwind related acquisitions of established firms that possess or produce sensitive technology.

Significant efforts to establish, strengthen and refine investment screening mechanisms that began around 2017 are continuing in many advanced economies. Most jurisdictions now operate review mechanisms that allow them to assess whether inward investment proposals may be injurious to their essential security interests and take corresponding action, especially and increasingly transactions that involve the acquisition of advanced technologies or related assets. The first section of this chapter describes how these efforts cover specifically companies that develop advanced technology and AI technology or related applications.

Investment screening mechanisms are only one among several instruments at governments’ disposal to manage essential security risks associated with advanced technologies. While reforms, introduction of new investment screening mechanisms in an ever greater number of jurisdictions, and expansion of scopes of existing mechanisms are reducing these actors’ access to advanced technology by way of acquiring established companies, foreign financing of research in advanced economies and inward or outward exchange of researchers emerge as a popular avenue for some emerging economies to acquire such know-how.1

Governments begin to take the resulting risk of undesirable technology transfer seriously. Some have responded with specific policies or proposals to close gaps that screening mechanisms may have left, while many others are still assessing needs, options and the balance of costs and benefits of any regulatory intervention.2 The second section of this chapter sets out the traits of these approaches that apply specifically to research into AI and similar advanced technology. These pioneering policies may foreshadow future and broader regulation to manage the proliferation of knowledge, rather than merely the transfer of firms that encapsulate such knowledge.

While investment screening mechanisms and related interventions can be effective in preventing the proliferation of advanced technology such as AI to malicious acquirers, they may, if not designed and implemented carefully, disturb the ecosystem that enables the development of such technology in the first place. This ecosystem thrives on openness, not least on openness to foreign capital, and it thrives on opportunities, entrepreneurship and market forces that drive innovation.3 While government intervention in this ecosystem is legitimate to safeguard essential security interests, policies and administrative practices need to strike a fine balance to minimise the repercussions on the economic environment for the development of technology – ultimately the foundation on which the potential of AI is built.

This quest to balance openness with imperatives to protect essential security interests has characterised investment policies for decades. Governments at the OECD developed policy principles to achieve this balance in 2009: the Guidelines for Recipient Country Investment Policies relating to National Security (2009 OECD Guidelines). These Guidelines have stood the test of time well, despite significant changes in the geopolitical and geo-economic environment since their adoption and the strong growth of investment from less-than-transparent jurisdictions and state-controlled entities as well as the more assertive stance of some economies. The third section concludes by setting out how these policies could inspire policy principles beyond screening of acquisitions of established enterprises in order to manage risk without stifling opportunities.

Key messages

Governments are managing risk for their essential security interests that may result from transfer of advanced technology to potentially malicious actors or hostile governments. Recent reforms in many advanced economies have focussed on screening of foreign investment in companies as one of the most prominent transmission channel for technology and knowledge transfer. Almost all advanced economies now cover advanced technology, including AI, by the scope of investment screening mechanisms.

As this avenue for technology acquisition is perceived to get narrower at least for some acquirers, foreign research funding and researcher exchanges are increasingly used to gain access to know-how that is not available domestically. Governments begin to take measures to address this gap in their defences but have to balance interests carefully to avoid stifling the conditions and the openness in which advanced technology development thrives.

Policy guidance developed at the OECD for investment policies related to national security in 2009 is an important reference point for policy-making for international investment in established enterprises. While there are important structural differences between international investment in established enterprises and research funding and cooperation, this policy guidance may provide some inspiration for policy principles for this latter area.

6.2. Managing essential security interests related to foreign acquisitions of AI assets in context

Policies that seek to safeguard essential security interests through screening of foreign acquisitions of sensitive assets have developed with a diverse and growing set of exposures in mind. When such mechanisms were first established in the context of the world wars of the first half of the 20^th century, predominant concerns related to espionage, sabotage and the intentional withholding of supply of defence goods.

The design of related policies and sectoral coverage of these early mechanisms – typically single-sector mechanisms enshrined in legislation governing the respective sector such as defence – testify to this focus of concerns. Where such policies were in place in the first half of the 20^th century, they typically focused on sensitive real estate and defence industries. Both sectors have retained their relative dominance on governments’ priority lists in this policy area until the present day (Figure 6.1).

With the privatisation of infrastructure, which many OECD Member countries started in the 1970s, and with foreign investment in this sector becoming possible, critical infrastructure was progressively included under the scope of investment review mechanisms to manage the risks associated with such foreign ownership. The concerns associated with foreign ownership of these assets were still centred on the risk of sabotage or intentional withholding of such sensitive assets.

Figure 6.1. Sector coverage of acquisition- and ownership-related policies to safeguard essential security interests in OECD Member countries (1960-2021)

Evolution of relative importance of individual sectors

Note: Coloured lines in the graph indicate the proportion of OECD Member countries whose investment review mechanisms in force in any given year cover the indicated sector, thus showing the relative importance attached to individual sectors in the context of investment screening for essential security interests. The basis for calculating 100% reflects only those of today’s 38 OECD Members that operated at least one review mechanism in that year.

6.2.1. New vulnerabilities emerge from international investment in advanced technology

An additional set of risks for essential security interests emerged around the same time. They include concerns about dependency on foreign-controlled suppliers, especially for defence procurement, and that the acquisition of advanced technology could level a technological edge, especially in defence applications. In the earlier decades of the Cold War period, export controls were used to limit access to advanced defence technology, but with the expansion of international investment, the role of acquisitions for technology proliferation across borders grew. Technological advances, specifically Japan’s success in the development of semiconductors in the 1970s and 1980s, among other developments, sparked concerns in the United States that its advantage in this field would shrink and that the United States would become dependent on foreign-owned suppliers of semiconductors and other technology with military applications. In response, the United States in 1975 laid the foundations of what would become the first modern investment review mechanism specifically established to manage national security risk (Graham and Marchick, 2006, p. 1[1]; Jackson, 2018[2]; United States Government Accountability Office, 2018[3]; Jackson, 2020[4])).

That technology transfer through international investment could undercut technological advantages and thus jeopardise essential security interests initially remained a concern essentially for defence applications proper. They led additional governments to introduce investment review mechanisms, most of which were initially focused on and limited to defence industries.

That focus on traditional defence industries began to fade only very recently, and it has only been in the past few years that advanced technology more broadly was included into the scope of investment review mechanisms.4 As documented in Figure 6.1, the inclusion of advanced technologies like AI into the scope of investment screening mechanisms only accelerated markedly in 2017, and only since 2019, more than half of OECD Members that screen foreign investment for essential security risks consider the implications of advanced technologies for such security interests.

The recognition of the role of AI as a foundational technology with dual-use applications – civil and military – plays an important part in this change of attitudes and policy: In 2021, twelve OECD Members had explicitly included AI in the definition of scope of their investment screening rules; not a single OECD Member had explicitly mentioned AI in this context before 2017 (Box 6.1).

Box 6.1. Inclusion of AI in the scope of investment screening mechanisms

The inclusion of AI in the scope and context of investment screening mechanisms first appeared in the United States legislation that reformed their earlier investment review mechanisms. This legislation is administered by the interagency Committee on Foreign Investment in the United States, better known under its acronym CFIUS. The reform legislation, the Foreign Investment Risk Review Modernization Act (FIRRMA), was enacted in 2018 and implemented over the subsequent two years, established particular procedures applicable to non-controlling investments in United States businesses that produce, design, test, manufacture, fabricate, or develop one or more critical technologies. Critical technologies includes, but is not limited to, emerging and foundational technologies controlled under section 1758 of the Export Control Reform Act of 2018 (50 U.S.C. 4817).

In Europe, AI was explicitly mentioned in the initial proposal of 2017 to establish EU-wide rules on investment screening which were later adopted as the Regulation 2019/452 of the European Parliament and of the Council of 19 March 2019 establishing a framework for the screening of foreign direct investments into the Union. Several EU Member States added AI to the scope of their screening mechanisms since (France in 2018, Italy 2020, Austria 2020, Slovenia 2020, Spain 2020, Germany 2021) or have planned its inclusion in ongoing reform efforts (e.g. Romania and Czech Republic). AI is also included in the list of ‘key enabling technologies’ of European Union interest under the Regulation 2019/452.

Other countries have taken action as well. Canada, in its 2021 update of the Guidelines on the National Security Review of Investments, includes AI in the list of sensitive technology areas. Australia introduced notification obligations in 2021 for foreign investments into businesses that develop, manufacture or supply critical technologies (including AI) for, or intended for, military or intelligence use; and encourages voluntary notification of similar investments in technologies with a civilian or dual use purpose. The United Kingdom included AI in the proposed list of critical technologies on which it consulted in late 2020.

OECD Members are not alone in their concerns about AI technology being acquired by foreign firms through international investment and acquisitions: As part of the update of its Catalogue of Technologies whose export is prohibited or restricted in August 2020, P.R. China included AI into the catalogue.

Source: (OECD, 2020[5]; OECD, 2021[6]).

Further changes to investment screening rules and policies have been made, if not specifically for AI acquisitions, but at least with regard to some of the specificities of AI acquisitions. Traditional designs of investment screening mechanisms were made with larger, publicly-listed companies in mind; only acquisitions of larger target companies, so the underlying rationale, could present meaningful risks for essential national security.

Companies that develop AI applications or related advanced technologies do not necessarily fit these criteria: In these sectors, many advances are made by small companies that may be held in private equity, that are not household names and that are not particularly visible to authorities.

In recognition that acquisitions of such smaller, non-listed companies may not come to the attention of investment screening authorities, several countries have adjusted their screening criteria and procedures to better capture such acquisitions of companies that develop advanced technology. For example, monetary or similar thresholds were abolished where these applied to investment screening on essential security grounds (United Kingdom in 2020, Australia in 2021, New Zealand in 2021), and notification requirements were introduced for sectors such as advanced technology where transactions might otherwise escape governments’ attention. Most recently, governments have also addressed loopholes of corporate governance arrangements that would give minority shareholders disproportionate rights to access of sensitive information or decision making, issues that are more likely to be observed in small companies that hold sensitive information (OECD, 2021[6]).

Enabling technologies for AI applications, such as semiconductors and quantum computing, have likewise been explicitly included into the scope of investment screening mechanisms in the past few years. Greater control over acquisitions of these technologies through international investments complement policies specifically focused on AI.

6.2.2. Greater scrutiny has not ended the international investment boom in AI

The inclusion of advanced technology and specifically AI under the scope of investment screening mechanisms in recent years has not left a mark in investment flow data in these areas or led to a significant number of visible government interventions against proposed transactions – at least not yet. Data availability is lagged, so trends and effects of most recent policy measures may only show in the future.

Inward investment into advanced economies in AI-related companies grew very significantly in aggregate value and, to a lesser extent, in numbers of transactions between 2015 and 2019 (Arnold, Rahkovsky and Huang, 2020, p. 14[7]) – against the trend of declining FDI flow volumes in that period overall (OECD, 2021[8]).

Caseloads under investment screening mechanisms also increased significantly in several countries over past years (Figure 6.2). However, few transactions are known to have been prohibited or unwound in OECD countries in recent years overall, and where such measures were taken, AI as such did not appear to have played a major role. Naturally, these data have to be used with caution: Investors may have withdrawn or not envisaged specific transactions where they sensed potential regulatory obstacles, hence this data may underestimate the effect on policy. However, the fact that transactions with specific other characteristics experienced increased and public scrutiny, it would appear that foreign investment in companies that develop AI or AI-applications suggests that new policies related to AI have been implemented with restraint.

Figure 6.2. Caseload under investment screening mechanisms in selected countries (2009-2021)

Note: Time-series shown where official data is made available by governments by late August 2021. The indicators shown depend on data availability and are not comparable across jurisdictions. Data for 2021 for Germany represents cases recorded by May 2021.

Source: OECD Secretariat calculations based on data reported by governments.

Judging by these preliminary metrics, investment screening mechanisms have likely had at most a minor impact on AI-related international investment.

Effects on a third potential outcome of more stringent investment screening cannot be assessed at this stage: To what extent will new investment, in particular greenfield investment in mobile, research-intensive sectors with security implications be allocated in jurisdictions that refrain from implementing controls over foreign investment. Investors in these areas may engage in regulatory arbitrage to avoid later security-related restrictions on their divestments (Pohl and Rosselot, 2020, p. 100[9]). Whether and to what extent such considerations influence the allocation of capital has not been subject to economic analysis.

6.3. Foreign investment in research: a new challenge calling for an adequate solution

While international investment in equity remains a very significant transmission channel for technology transfer, other avenues exist and develop in parallel. Funding of research in foreign countries, and inward and outward researcher exchanges specifically, are a significant legal5 avenue through which such transfer of know-how across borders can take place (Hannas and Chang, 2019[10]).

Joint research work in universities or research institutions, or research financed by foreign governments or foreign enterprises, or inward or outward researcher exchanges allow these researchers and their principals or funders to tap into knowledge, know-how and networks to acquire capabilities that are not available domestically.

These forms of allocation of resources to acquire know-how rather than investing in enterprises can substitute for equity investments in the foreign market, especially where and when such equity acquisitions raise suspicion with local authorities or are difficult to execute. Investment in know-how acquisitions may become more attractive as perceptions about more stringent investment screening specifically for AI and related advanced technologies may dampen the appetite for equity acquisitions.

International research funding and researcher exchanges are regulated under rules that are different from investment screening. Some aspects of technology transfer are subject to export control regimes,6 but these rules do not cover all means of carrying know-how across borders and, especially for foundational research with broad, yet to define applications, they may not always be a suitable instrument to address certain aspects of technology transfers.7

International collaboration and cooperation in research is widely recognised as critically important for advancing science and technology as well as solving global challenges.8 In many OECD countries, independence of research institutions, and, especially in federal states, split competencies to regulate the fields of research and national security, further complicate government action in this area. Furthermore, the implantation of foreign-funded research in one’s territory and the attraction of foreign talent and university fees are perceived as an opportunity to generate high-paying employment, advance research in one’s institution, and is correspondingly generally welcome.

The implications of legal technology transfer through cross-border research funding and researcher exchanges have come under greater scrutiny lately,9 in particular where this cooperation concerns advanced technology and specifically AI.10 It has been suggested that some of the research funding and researcher exchanges are systematic attempts to extract know-how for subsequent use in defence applications developed abroad (Brown and Singh, 2018[11]; Joske, 2018[12]; Silcoff et al., 2018[13]; Segal and Gerstel, 2019[14]; National Science Foundation, 2019[15]; United States Senate Permanent Subcommittee on Investigations, 2019[16]; Hannas and Chang, 2019[10]; Lloyd-Damnjanovic and Bowe, 2020[17]) and have argued that a policy responses are warranted (JASON, 2019[18]; Kratz et al., 2020[19]).

As governments became more suspicious of foreign research involvement in sensitive areas, and as the volume of exchanges grew, especially in these areas, some began to recognise the issue11 and to consider restrictions on foreign researchers (Edwards, 2016[20]) or on foreign funding of research and the transfer of results abroad. The United States complemented its export control rules in 2018 to manage technology transfer,12 put in place a targeted limitation of entry for certain students,13 and legislation was under consideration in June 2021 that aims at managing national security implications of foreign research funding and researcher exchanges.14 Australia has passed legislation in 202015 that requires notifications about non-commercial arrangements that any subnational entities, including research institutions, have concluded or plan to conclude with foreign governments. Canada issued a Research Security Policy Statement16 in March 2021 that aims at managing the risk of foreign funding arrangements for its national security. Japan is reported to have announced in April 2021 that it would require universities to disclosure foreign financial contributions when applying to government funds in order to avoid transfer of research that could be used for military purposes (Oikawa, 2021[21]).

By June 2021, the issue of transfer of advanced technology and AI-related technologies through foreign research funding and international researcher exchanges had become a broadly shared policy priority among advanced economies as documented by its inclusion in the EU-US Summit of 15 June 2021.17

6.4. Managing the implied risks of openness without forgoing benefits

There are parallels between international investment in enterprises and foreign funding of the generation of knowledge:

• Openness brings benefits to home and host societies and fosters prosperity and innovation, and both are keenly needed to address today’s and tomorrow’s challenges;

• Openness may, in both areas, occasionally bring risk for essential security interests that warrant policy intervention; and

• Diligent calibration of such policy intervention is crucial to avoid damages to the ecosystem in which international investment and international research cooperation thrive and that are required to generate the associated benefits.

In the area of international investment, advanced economies took guidance from time-tested policy principles when designing policies to manage security risks without forgoing the benefits of openness: the 2009 OECD Guidelines for Recipient Country Investment Policies relating to National Security. These Guidelines establish specific standards regarding non-discrimination, transparency, predictability, regulatory proportionality and accountability to allow societies to benefit from open investment environments while managing occasional risks that this openness can bring.

As governments are beginning to address security implications associated with international research cooperation and researcher exchanges, they may wish to consider which of these principles, if not all, could usefully inspire the design of rules in this area.

For example, the section of the 2009 OECD Guidelines on “proportionality” calls on governments to design investment policies so that “restrictions on investment, or conditions on transaction, should not be greater than needed to protect national security and they should be avoided when other existing measures are adequate and appropriate to address a national security concern”. Very similar aspects have been identified for research cooperation, especially in areas of emerging technologies and AI: Broad prohibitions are identified as too blunt, risk stifling innovation and progress altogether, or push research and development abroad to more permissive countries with even greater detriment for essential security interests and innovation at home (United States Intelligence, National Security Commission on Artificial Intelligence, 2021, p. 176[22]; Williams, 2018[23]; National Research Council, 2007, p. 27[24]).

The 2009 OECD Guidelines also call for transparency and predictability of policies designed to safeguard essential security interests. They specifically emphasise the need to make regulatory objectives and practices as transparent as possible so as to increase the predictability of outcomes. Research and innovation often require significant upfront investment and personal commitments of researchers; also, many projects run over longer periods, findings may lead them into directions that were not initially anticipated, and produce results or applications that were not planned. This is all the more the case in a resource intensive, fast-moving and foundational field like AI, for which human ingenuity may find applications that we cannot foresee today. Understanding rules, policies and concerns are crucial conditions to create trust that certain research can be successfully carried out in a given jurisdiction – much like in the field of long term commitment of assets in foreign investment.

Finally, the 2009 OECD Guidelines call for accountability of policy actions to citizens on whose behalf these measures are taken, achieved through oversight, public reporting, regulatory impact assessment and the like. Again here, the parallel between international investment in equity and foreign funding of knowledge generation is striking: Accountability generates trust, avoids overreach, and is the foundation for legitimacy that is needed as much for international investment in enterprises as it is for the allocation of resources to research.18

Most advances in science are achieved on the basis of or in analogy to preceding findings of others that are assessed, further developed and refined. Policy makers could take inspiration from this incremental approach and look at the advances their peer policy makers have made in related fields. Addressing the risks that international research cooperation in AI and other advanced technology may occasionally generate, be it through foreign funding or inward and outward researcher exchanges, could take inspiration from what makes research so frighteningly successful. The time-tested principles developed for international investment may have some insights in store for international research cooperation, too.

References

[7] Arnold, Z., I. Rahkovsky and T. Huang (2020), Tracking AI Investment: Initial Findings From the Private Market, Center for Security and Emerging Technology, http://dx.doi.org/10.51593/20190011.

[11] Brown, M. and P. Singh (2018), China’s Technology Transfer Strategy:How Chinese Investments in Emerging Technology Enable A Strategic Competitor to Access the Crown Jewels of U.S. Innovation, Defence Innivation Unit Experimental, https://admin.govexec.com/media/diux_chinatechnologytransferstudy_jan_2018_(1).pdf.

[27] d’Hooghe, I. and J. Lammertink (2020), Towards Sustainable Europe-China Collaborationin Higher Education in Research, https://leidenasiacentre.nl/wp-content/uploads/2020/10/Towards-Sustainable-Europe-China-Collaboration-in-Higher-Education-and-Research.pdf.

[20] Edwards, J. (2016), U.S. targets spying threat on campus with proposed research clampdown, Reuters, https://www.reuters.com/article/us-usa-security-students-idUSKCN0YB1QT.

[25] Federal Office for Economic Affairs and Export Control (BAFA) (2019), Export Control and Academia Manual, https://www.bafa.de/SharedDocs/Downloads/EN/Foreign_Trade/ec_academia.pdf.

[1] Graham, E. and D. Marchick (2006), U.S. National Security and Foreign Direct Investment, Economics, Peterson Institute for International Economics, https://www.piie.com/bookstore/us-national-security-and-foreign-direct-investment.

[10] Hannas, W. and H. Chang (2019), China’s Access to Foreign AI Technology, https://cset.georgetown.edu/wp-content/uploads/CSET_China_Access_To_Foreign_AI_Technology.pdf.

[4] Jackson, J. (2020), The Committee on Foreign Investment in the United States (CFIUS), https://crsreports.congress.gov/product/pdf/RL/RL33388/93.

[2] Jackson, J. (2018), The Committee on Foreign Investment in the United States (CFIUS), Congressional Research Service, https://crsreports.congress.gov/product/pdf/RL/RL33388/68.

[18] JASON (2019), Fundamental Research Security, https://nsf.gov/news/special_reports/jasonsecurity/JSR-19-2IFundamentalResearchSecurity_12062019FINAL.pdf.

[12] Joske, A. (2018), Picking flowers, making honey – The Chinese military’s collaboration with foreign universities, Australian Strategic Policy Institute, https://s3-ap-southeast-2.amazonaws.com/ad-aspi/2018-10/Picking flowers%2C making honey_0.pdf.

[19] Kratz, A. et al. (2020), Chinese FDI in Europe: 2019 Update – Special Topic: Research Collaborations, Rhodium Group; Merics, https://rhg.com/wp-content/uploads/2020/04/MERICS-Rhodium-Group_COFDI-Update-2020-2.pdf.

[17] Lloyd-Damnjanovic, A. and A. Bowe (2020), Overseas Chinese Students and Scholars in China’s Drive for Innovation, https://www.uscc.gov/sites/default/files/2020-10/Overseas_Chinese_Students_and_Scholars_in_Chinas_Drive_for_Innovation.pdf.

[26] National Academy of Engineering (1982), Scientific Communication and National Security, National Academies Press, Washington, D.C., http://dx.doi.org/10.17226/253.

[24] National Research Council (2007), Science and Security in a Post 9/11 World: A Report Based on Regional Discussions Between the Science and Security Communities, National Academies Press, Washington, D.C., http://dx.doi.org/10.17226/12013.

[15] National Science Foundation (2019), Personnel Policy on Foreign Government Talent Recruitment Programs, https://www.nsf.gov/bfa/dias/policy/researchprotection/PersonnelPolicyForeignGovTalentRecruitment Programs07_11_2019.pdf.

[8] OECD (2021), FDI in Figures, https://www.oecd.org/investment/FDI-in-Figures-April-2021.pdf.

[6] OECD (2021), Investment policy developments in 62 economies between 16 October 2020 and 15 March 2021, https://www.oecd.org/daf/inv/investment-policy/Investment-policy-monitoring-March-2021-ENG.pdf.

[28] OECD (2021), Transparency, Predictability and Accountability for investment screening mechanisms, Research note by the OECD Secretariat, https://www.oecd.org/daf/inv/investment-policy/2009-Guidelines-webinar-May-2021-background-note.pdf.

[5] OECD (2020), Inventory of investment measures taken between 16 September 2019 and 15 October 2020, https://www.oecd.org/daf/inv/investment-policy/FOI-investment-measure-monitoring-October-2020.pdf.

[21] Oikawa, A. (2021), Japan tightens rules on tech theft, https://asia.nikkei.com/Business/Technology/Japan-tightens-rules-on-tech-theft-to-safeguard-research-with-US.

[9] Pohl, J. and N. Rosselot (2020), Acquisition-and ownership-related policies to safeguard essential security interests - Current and emerging trends, observed designs, and policy practice in 62 economies, https://www.oecd.org/investment/OECD-Acquisition-ownership-policies-security-May2020.pdf.

[14] Segal, S. and D. Gerstel (2019), Research Collaboration in an Era of Strategic Competition, Center for Strategic and International Studies, https://csis-website-prod.s3.amazonaws.com/s3fs-public/publication/190925_Segal%26Gerstel_ResearchCollaboration.pdf.

[13] Silcoff, S. et al. (2018), How Canadian money and research are helping China become a global telecom superpower, https://www.theglobeandmail.com/canada/article-how-canadian-money-and-research-are-helping-china-become-a-global/.

[3] United States Government Accountability Office (2018), Committee on Foreign Investment in the United States - Treasury Should Coordinate Assessments of Resources Needed to Address Increased Workload, https://www.gao.gov/assets/gao-18-249.pdf.

[22] United States Intelligence, National Security Commission on Artificial Intelligence (2021), Final Report, https://www.nscai.gov/wp-content/uploads/2021/03/Full-Report-Digital-1.pdf.

[16] United States Senate Permanent Subcommittee on Investigations (2019), Threats to the U.S. Research Enterprise: China’s Talent Recruitment Plans, https://www.hsgac.senate.gov/imo/media/doc/2019-11-18%20PSI%20Staff%20Report%20-%20China%27s%20Talent%20Recruitment%20Plans.pdf.

[23] Williams, R. (2018), “In the Balance: The Future of America’s National Security and Innovation Ecosystem”, Lawfare, https://www.lawfareblog.com/balance-future-americas-national-security-and-innovation-ecosystem.

Notes

← 1.  This chapter does not deal with non-commercial forms of technology acquisition, such as theft of intellectual property or espionage.

← 2.  The OECD Recommendation of the Council on International Co-operation in Science and Technology, initially adopted in 1988 but revised in 2021, calls on Adherents to remove barriers to mutually beneficial international co-operation in science and technology and offers recommendations to expand such cooperation with a view to contribute to sustainable development, inclusive economic growth and social well-being.

← 3.  The OECD Recommendation of the Council on Artificial Intelligence (2019) emphasises the various aspects that governments wish to keep in mind to provide an enabling ecosystem.

← 4.  See on the rationale specifically for AI (United States Intelligence, National Security Commission on Artificial Intelligence, 2021, p. 12[22]).

← 5.  Illegal means to transfer technology, in particular theft and espionage are not considered in the context of this note.

← 6.  Corresponding guidance is in place for example in Japan (Ministry of Economy, Trade and Industry, Trade Control Department: “Guidance for the Control of Sensitive Technologies for Security Export for Academic and Research Institutions”, 3rd Edition, October 2017) and Germany (Federal Office for Economic Affairs and Export Control (BAFA), 2019[25]).

← 7.  The German government states that “basic scientific research is not subject to export controls”. (Federal Office for Economic Affairs and Export Control (BAFA), 2019, p. 54[25]).

← 8.  See the Preamble of the Revised Recommendation of the Council on International Co-operation in Science and Technology, adopted on 23 June 2021.

← 9.  Similar concerns had been raised in the 1980s in the United States, then in relation to technology acquisition by Eastern Bloc nations (National Academy of Engineering, 1982[26]). They led, among others to the adoption of the National Security Decision Directive 189 (NSDD-189) on 21 September 1985.

← 10.  A vivid debate about potential foreign interference and intimidation in research institutions is taking place contemporaneously in several countries, often the same that have expressed greatest concern about the implications of researcher exchanges and foreign research funding. Alleged censorship of certain social media content has also been cited in relation to foreign investment reviews. These issues raise other aspects than those related to transfer of technology and are thus not further discussed here. Context and other contemporaneous concerns are summarised in (d’Hooghe and Lammertink, 2020, p. 59[27]).

← 11.  E.g. Government bill 2020/21:60 Research, freedom, future – Knowledge and innovation for Sweden, 17 December 2020, in particular section 17.3 of the explanatory memorandum.

← 12.  Export Control Reform Act of 2018 (ECRA; P.L. 115-232, Subtitle B, Part I)

← 13.  “Suspension of Entry as Nonimmigrants of Certain Students and Researchers From the People’s Republic of China”, Proclamation by the President of the United States of America 10043 of May 29, 2020, Federal Register Vol. 85, No. 108, June 4, 2020. “Technology Alert Lists” (TAL), of which older, publicly available versions specifically mention AI, are used by consular officers to screen applicants for United States visa.

← 14.  Bill S.1260 — 117th Congress (2021-2022).

← 15.  Australia’s Foreign Relations (State and Territory Arrangements) Act 2020.

← 16.  Research Security Policy Statement – Spring 2021, 24 March 2021.

← 17.  EU-US Summit Statement, “Towards a renewed Transatlantic partnership”, 15 June 2021.

← 18.  More details on the implementation of the 2009 OECD Guidelines is available in (OECD, 2021[28])