The European Union’s AI Governance Agenda: Ethical Investment and Flexible Regulatory Approaches
Abstract
This paper examines the European Union’s (EU) approach to artificial intelligence (AI) governance, focusing on the period from the publication of Ethics Guidelines for Trustworthy AI (2019) onwards. It explores the EU’s strategy to address its “digital gap” in AI development and adoption. The study employs a theoretical framework synthesizing concepts of AI policy as a “horizontal area” (Niklas and Dencik 2020), “tentative governance” (Kuhlmann and Rip’s 2018), and an “ethics first” approach (Floridi 2018). The analysis reveals and discusses the EU’s distinctive prioritization of ethical considerations in AI development, manifested through substantial public investments and flexible regulatory interventions.
Introduction and Theoretical Framework
“Europe’s digital gap”: this is how analysts at the McKinsey Global Institute in 2019 described Europe’s lagging behind global leaders such as the USA in the realm of AI development and adoption. They cautioned that while digitization is essential for the advancement of AI, Europe continues to lag approximately 35% behind the United States; a trend that has not diminished in recent years. More recently, the State of the Digital Decade 2024 report highlights both progress and challenges in Europe’s digital transformation, including in AI adoption. While there has been advancement in areas such as 5G coverage and digital skills, the report notes that the percentage of enterprises using AI increased by a mere 0.4% in 2 years, from 7.6% in 2021 to 8.0% in 2023, significantly below the 75% target for 2030 (see State of the Digital Decade Report, 2024, Annex 2). The analysis emphasizes the need for European Union (EU) countries to strengthen their actions and be more ambitious in achieving the Digital Decade’s goals, in order to ensure EU’s future economic prosperity.
Europe’s technological divide is not only evident in the slow adoption of AI technologies but also in the stark disparities in AI investment between Europe and global leaders, like North America. A European Commission’s (EC) 2018 report illustrates the severity of this investment disparity, revealing that private AI investments in Europe totalled only €2.4-3.2 billion, compared to a much higher €12.1-18.6 billion in North America. This significant difference in investment underscores Europe’s struggle to foster a competitive AI ecosystem, one capable of stimulating innovation at the same pace as its counterparts. Scholars, such as Mazzucato (2017) caution about the impact of chronic underinvestment in high-risk, early-stage research, emphasizing the importance of public investment in fostering innovation, and warning that underfunding in these critical areas can lead to stagnation in technological advancement and economic growth. In the case of AI this gap is particularly concerning, because the transition from basic research to practical application can be extraordinarily short (Sepp Hochreiter, 2023), and many European enterprises still lag behind global competitors due to challenges in rapidly translating research into practical applications (State of the Digital Decade Report, 2024, Annex 2).
The persistent digital divide underscores the critical importance of pan-European AI policy initiatives in accelerating AI adoption. To analyse this complex landscape, the present discussion adopts a theoretical framework that synthesizes three key theoretical concepts from existing literature and policy approaches. First of all, this is Niklas and Dencik (2020) theorization of AI policy as a “horizontal area” that spans multiple domains and sectors. The scholars highlight the dual role of state intervention in this process, which encompasses both public funding for research and regulatory oversight for industry. Secondly, Kuhlmann and Rip’s (2018) concept of “tentative governance” refers to flexible and adaptive regulatory approaches needed to address the dynamic and rapidly evolving nature of AI technologies. Thirdly, the “tentative governance” policy concept is extended by drawing upon Floridi’s (2018) guidance about “ethics first” approach to AI development and governance.
The EU’s regulatory stance distinctively prioritizes ethical considerations in AI development and deployment, a principle firmly established by the 2019 Ethics Guidelines for Trustworthy AI, which have laid the foundation for subsequent AI policy frameworks. Thus, the subsequent brief chronological analysis will examine key European AI-related policies and programs from 2019 onwards through these three theoretical lenses, focusing on how these initiatives aim to bridge the digital divide while adhering to principles of responsible and ethical AI development through flexible and adaptive “tentative governance” approaches.
In accordance with the theoretical guidance about the dual role of state intervention in AI policy, the discussion will first explore public funding as a catalyst for ethical AI innovation, followed by an assessment of key recent EU AI policy interventions. This structure reflects the complementary nature of research funding and regulatory measures in shaping the European AI landscape.
Public Funding as Catalyst of AI innovation
In response to the challenge of the “digital gap,” Europe has increasingly relied on public funding and collaborative research initiatives as strategic tools to bridge this divide. Importantly, these funding initiatives integrate the ethics-first approach, ensuring that ethical considerations remain at the forefront of EU’s AI development strategy.
Horizon Europe, with its substantial budget of €93.5 billion for 2021-2027 (note that this total includes all areas of research and innovation, not just AI), serves as the centrepiece of this strategy, deliberately committing to “act as a synergetic force across the EU funding programmes.” Within this, a significant portion is allocated to digital technologies, including AI. For instance, the “Digital, Industry and Space” cluster, which encompasses AI research, has been allocated €15.3 billion. Public investment acts supports innovation in critical technologies, while also emphasizing the development of ethically aligned AI systems. For example, Horizon Europe’s Ethics Appraisal Process is designed to ensure compliance with ethical standards throughout EU-funded research projects and now includes a mandatory ethics appraisal procedure for proposals that raise ethical issues.
This approach is complemented by other initiatives within the European AI ecosystem. The Digital Europe Programme (DIGITAL), while separate from Horizon Europe, synergizes with it by focusing on building strategic digital capacities and deploying digital technologies. DIGITAL has an overall budget exceeding €7.9 billion for 2021-2027. The programmes will collectively invest €1 billion per year in AI, which totals €4 billion over four years (2021-2027). Both Horizon Europe and Digital Europe require ethics review procedures, even though they are more robustly defined within Horizon Europe compared to DIGITAL. This integrated funding strategy aims to establish a seamless continuum of support from foundational research to market deployment, with ethical considerations embedded throughout.
European initiatives emphasize not only funding for research projects but also support for training and mobility programs. For instance, the Marie Skłodowska-Curie Actions, with a budget of €6.6 billion under Horizon Europe, specifically targets researchers in AI and other cutting-edge fields, while also promoting ethical awareness in AI research. Additionally, the European Innovation Council (EIC) has allocated €10.1 billion to promote breakthrough innovations, including AI, from laboratory settings to market products, with a focus on responsible and ethical development.
Individual member states are also making substantial investments in AI. For example, the German Federal Government (2020) has pledged €3 billion for its AI strategy up to 2025, which includes a strong emphasis on “AI made in Europe” that adheres to European ethical standards. France’s National Strategy for AI (SNIA) had a budget of €1.5 billion in public funds for five years (2018-2022). The second phase of France’s national AI strategy (promulgated in 2021) dedicated a total of €2.22 billion for AI over the next 5 years, of which €1.5 billion in public funding and €506 million in private co-financing, with ethical AI development as a key priority.
Collectively, the investments discussed in this section represent a substantial commitment to AI advancement. In its Coordinated Plan on Artificial Intelligence Review (2021), the European Commission estimates that the combined public and private investments in AI in Europe could reach €20 billion per year over the next decade. This level of investment represents a significant step towards closing the AI investment gap and fostering a thriving AI ecosystem in Europe that prioritizes ethical considerations.
Regulatory Interventions
The ensuing brief overview will discuss key European AI-related policies and programs promulgated since 2019.
The analytical exposition begins with the publication of the Ethics Guidelines for Trustworthy AI by the High-Level Expert Group on AI. These guidelines laid the ethical foundation for Europe’s approach to AI development. An emphasis was placed on principles such as human agency, privacy, and non-discrimination, reflecting an “ethics-first” approach to AI governance. This initial step was of great importance in establishing the foundation for future policies. It signalled Europe’s commitment to developing AI systems that not only push technological boundaries but also align with fundamental European values. The guidelines has served as a compass, guiding subsequent regulatory efforts and shaping the European vision of “trustworthy AI.”
Recognizing that AI development is intrinsically linked to data availability, the EU launched the European Strategy for Data in 2020. This strategy aimed to create a single market for data and envisioned a future where data flows freely within the EU while adhering to strict privacy and data protection rules. Moreover, it sought to democratize AI development by ensuring fair access to data, preventing the concentration of power in the hands of a few large tech companies.
In 2021, the EU launched the Digital Decade policy program, setting ambitious targets for Europe’s digital transformation by 2030. This program recognized AI as a key technology for achieving these goals and emphasized creating a “secure and performant sustainable digital infrastructure.” The Digital Decade program marked a shift from isolated policies to a more comprehensive vision of Europe’s digital future. It contextualized AI within broader digital transformation goals, highlighting the interconnectedness of various technological advancements and societal changes.
Proposed in 2021, the AI Act is a landmark legislation that would become the cornerstone of European AI regulation. As analysed by Veale and Zuiderveen (2021), it adopts a risk-based approach, categorizing AI systems based on their potential societal impact and implications for individual rights. This nuanced approach allowed for stringent oversight of high-risk applications while fostering innovation in lower-risk areas. Therefore, the act exemplifies the concept of tentative governance, providing a flexible framework that could adapt to the rapid pace of AI development.
In 2022, the EU introduced the Digital Services Act and the Digital Markets Act. While not AI-specific, these acts played a crucial role in shaping the regulatory environment for AI technologies in Europe. They set rules for online platforms and digital markets, indirectly influencing the development and deployment of AI technologies. These acts demonstrate tentative governance through their layered approach to regulation, with different rules applying to different sizes and types of digital platforms, allowing for adaptability as the digital landscape evolves.
Running parallel to these regulatory efforts is Horizon Europe, the EU’s key research and innovation program for 2021-2027. While not exclusively focused on AI, Horizon Europe provides significant funding for AI research and development, fostering collaboration between academia, industry, and the public sector. Horizon Europe embodies tentative governance through its adaptive funding mechanisms and focus areas, which can be adjusted to address emerging AI challenges and opportunities.
As seen from this brief exposition, from the ethical guidelines of 2019 to the risk-based approach of the AI Act and the adaptive research funding of Horizon Europe, there is a consistent pattern of flexible, adaptive governance strategies, which prioritise ethical considerations. These strategies allow the EU to navigate the complex and uncertain landscape of AI development while maintaining a balance between technological innovation and ethical considerations.
Conclusions
The EU must proactively harness the potential of AI to prevent becoming merely a consumer of solutions developed elsewhere and to capitalize on the transformative opportunities this technology presents. The EU’s approach to AI governance combines increased public funding with increasingly more flexible and adaptive regulatory frameworks. It aims not only to close the AI gap with global leaders but to establish Europe as a pioneer in human-centric, trustworthy AI that balances innovation with ethical considerations and societal needs.
