The EU Competitiveness Compass and AI regulation: competitive edge, ethical core

EU funding to encourage and support industrial adoption of AI in key sectors 

A key pillar of the Compass Strategy is closing the innovation gap with the US and China, with a view to establishing the EU as a global leader in the race towards innovation in tech sectors, particularly AI. The Compass Strategy echoes the words of President Ursula von der Leyen, who recently announced her ambitious plan to "turn Europe into the ‘AI continent’".⁶ 

The Compass Strategy outlines several initiatives, such as the AI factories project⁷ and the EU Cloud and AI Development Act. These are supposed to increase the supercomputing power of the EU, create the necessary data infrastructures and give companies access to cloud infrastructures and services, with the ultimate goal of building a thriving ecosystem to train advanced AI models and develop AI solutions. In addition, the EU’s goal is to increase the adoption and integration of AI in strategic industrial sectors, such as manufacturing, automotive, energy, robotics or pharmaceutical.

AI investments demand significant funding from private companies, much of which would be challenging to secure without financial support by Member States (i.e. State aid) and the EU. Conscious of this, the EU has announced several measures and packages that complement each other, aimed at supporting AI startups and SMEs, such as the Digital Europe Programme (DIGITAL),⁸ an EU funding programme aimed to encourage investments in digital technology and infrastructure. The EU also plans to support the development and implementation of generative AI through Horizon Europe,⁹ a funding programme for research and innovation, and the Digital Europe Programme,¹⁰ which supports programmes aimed at contributing to digital transformation.

EU-funded projects are already underway, such as the OpenLLM 🇫🇷 🇪🇺, which is building the first family of open-source Large Language Models covering all official EU languages.¹¹

State aid to encourage AI investments in the EU

The Compass Strategy, clearly sets out the goal for the EU to “lead in the global economy of tomorrow”.¹² This “EU first” approach is echoed throughout the document (with mentions, for example, to the preferential treatment in public procurement). Together with the focus on AI funding, it will pave the way for greater opportunities for Member States to grant public funding for AI projects, especially if they can demonstrate that in the absence of such funding companies would locate their investments outside the EU.

For example, to enable Member States to attract strategic investments, the EC’s State Aid rules for Research and Development and Innovation¹³ (which explicitly cover funding for AI projects), provide the possibility to declare State aid compatible with the internal market on the basis of the so-called "matching clause". The idea underlying the matching clause is essentially that Member States can grant funding to companies to invest in the EU, in order to match the amount of State funding they could have received if they had chosen to carry out their project in a third country. The subsidies race between EU and third countries targets in particular China and USA (jurisdictions which highly subsidize innovative technologies) and has already led Member States to grant record State aid in innovative sectors (as, for instance, the EUR 10 billion subsidy given by Germany to American chip maker Intel¹⁴).

Expanding funding opportunities, however, is only one side of the solution. The Compass Strategy recognizes that red tape remains a major concern for investors and that State aid screening must be modified to be less burdensome. This is aligned with Commissioner Teresa Ribera’s announcement that simplifying and speeding up the screening of State aid is one of her priorities.¹⁵ Hopefully the EU will take measures to put these plans into action. 

Further coordination to reduce barriers to free movement within the EU market

The Compass Strategy outlines further coordination efforts between Member States and the EU, in particular in emerging areas such as AI, to avoid fragmentation of the EU Internal market as a result of diverging national regulatory frameworks. A first step has already been taken by the adoption of the AI Act,¹⁶ which aims at removing obstacles for cross-border AI single market that would threaten the EU’s digital sovereignty.¹⁷ The Compass Strategy announced further steps in this direction, by announcing the implementation of an Apply AI Strategy and Data Union Strategies by Q3 2025.

The EC also plans on proposing a "Competitiveness Coordination Tool" to foster coordination between Member States and set common competitiveness priorities in key areas, notably digital infrastructure and AI vertical use cases, as well as outline a common EU governance mechanism for industrial and investment policies. The EC’s plan is that the Competitiveness Coordination Tool would apply to State aid granted by Member States but also EU funding and private capital.

While the Compass Strategy fundamentally aims to enhance European economic competitiveness, including by incentivizing investments in the EU, its underlying model reflects the historical European approach aimed at reconciling investments and innovation with high levels of protection for EU citizens and a balanced regulatory framework.¹⁸

AI Regulation in the EU: A Dual Approach of Legislation and Industry Self-Regulation 

Unlike more laissez-faire models emerging elsewhere, the EU's approach to AI regulation, embodied in the AI Act, admittedly aims to strike the right balance between fostering innovation while safeguarding fundamental rights. Rather than implementing sweeping restrictions, the AI Act adopts a granular, risk-based framework that focuses primarily on high-risk AI systems and practices that may threaten fundamental rights. In turn the AI Act maintains a more flexible approach to remaining applications. In other words, the EC’s regulatory choice is to dynamically govern AI development through targeted interventions aimed at ensuring that AI technologies are safe and compliant with EU laws and values, while avoiding overly prescriptive rules that would make the AI Act unnecessarily restrictive and not future-proof. 

A prime example of this approach are the guidelines on prohibited practices (Article 5 of the AI Act) issued by the EC on 4 February 2025. While providing practical guidance, these guidelines emphasize the need for case-by-case assessment rather than rigid rules, and maintain flexibility by allowing the EC to refine their interpretation over time.¹⁹

This balanced approach is also reflected in complementary initiatives like the AI Pact²⁰ – already signed by over 100 companies as a form of self-regulation – and in the broader efforts to coordinate AI standards through non-binding guidelines and principles.²¹

Less is more?

This broad EU framework on AI, mainly created by the AI Act and complemented by the EC's guidelines (as well as the additional guidance that will be issued in the forthcoming months) and the above self-regulation initiatives, doesn't view regulation as an obstacle, but rather as a vital mechanism to drive responsible technological advancement. 

While this is indeed commendable – and reflects, at its heart, the core principles of EU law as a whole – these initiatives must be read in conjunction with the compelling need for regulatory simplification, as expressed in the Compass Strategy. The same EC's guidelines on prohibited practices mentioned above raise questions on whether a 140-page guidance commenting on one single provision of the AI Act fits in such a "simplification" effort. 

That said, by acknowledging that the availability of large and high-quality data is an essential component of developing AI, the EC is for instance committed to proposing a Data Union Strategy to improve and facilitate secure private and public data sharing, by simplifying the existing regulatory regime, and accelerating the development of new systems or applications. The Compass Strategy announces the EC's Simplification Omnibus packages, aimed at streamlining sustainable finance regulations by creating more proportionate and flexible requirements to mobilize investments while reducing administrative complexity for businesses in transition.

Whether these efforts will produce the desired outcomes in terms of compliance or have the opposite effect of introducing new complexities for businesses remains uncertain. As part of a broader effort to strengthen the Single Market, the EC is also advancing harmonization measures to reduce legal fragmentation, ensuring a consistent regulatory environment that supports both competitiveness and legal certainty. While these initiatives signal a strong commitment to fostering AI leadership within a rules-based framework, their practical impact will ultimately depend on effective implementation, coordination across Member States, and the ability to strike the right balance between innovation and oversight.

The multi-layered nature of AI technologies: navigating overlaps and evolving legislation

In this context, the Compass Strategy sets out ambitious and well-intentioned goals, but the EU will have to carefully think through how these new initiatives fit into the broader legislative framework. Beyond the goal of merely "simplifying" existing EU legislation, it is imperative that the EC and EU legislature also take an active role in clarifying potential regulatory overlaps. 

While this aspect is not explicitly addressed in the Compass Strategy in relation to AI, it is nonetheless a responsibility assigned to the EC under Article 96 of the AI Act. Read in conjunction with the EC's commitment to "ensur[ing] a level playing field across the Single Market, as well as fight fragmentation and gold plating, [to] pursue a forceful approach to full harmonisation and enforcement"²², the AI Act will also be part of this fight against fragmentation, as national legislations are already on the rise.²³

Thus far, much of the debate has understandably focused on the General Data Protection Regulation ("GDPR") and the existential challenges when it comes to generative AI and data protection compliance, yet an equally pressing issue is the intersection between the AI Act and the Digital Services Act ("DSA"). Overlapping provisions in areas such as risk assessment, AI-integrated platforms, and transparency obligations for users create legal uncertainty and potential burdensome duplication of compliance efforts, requiring a simplified approach to these rules to achieve a coherent regulatory approach.

Similar concerns apply to the interplay between the Digital Markets Act ("DMA")²⁴ and AI. Although AI technologies (such as Language Learning Models ("LLMs") and generative AI), are not expressly included within the scope of the DMA, the High-Level Group of the DMA ("HLG") has clarified that, provided that AI systems are embedded into the designated core platform services, compliance with the DMA must be evaluated based on how AI systems influence or determine behaviors that fall under the provisions of this regulation.²⁵ HLG has also joined the discussion around the regulation of AI, admitting that it is becoming a central policy issue in the EU and globally and stating that they will follow developments in this area and explore the interactions between the DMA and other regulatory instruments, without having (yet) made a statement about the express inclusion of LLMs and generative AI within the scope of the DMA. 

Ultimately, ensuring greater legal clarity in these domains will be essential to fostering a competitive and innovation-friendly digital environment within the EU. Such clarity must reflect the cross-disciplinary nature of AI technologies, which by design transcend traditional regulatory boundaries.

The Compass Strategy outlines several legal initiatives aimed at increasing the EU’s ability to protect and boost its AI industry against unfair investments resulting in leakage of knowledge, and unfair trading conditions from third countries. The announced revision of the EU Procurement Directive by 2026, with its European-first requirement, the highlighted approaches to Foreign Direct Investment Screening, Export Controls, Outward Screening, and the Foreign Subsidy Regulation will have significant implications on the EU foreign trade policy. This is particularly true for strategic sectors, such as AI.  

Screening foreign investments in AI

Over the past few years there has been an increasing push by EU and Member States to expand the role of foreign direct investment (“FDI”) to prevent the outflow of know-how in critical technologies, such as AI. While FDI screening still falls within the competence of EU Member States, the EC has pushed for greater coordination at the EU level. The Compass Strategy reflects these trends, mirroring other EU initiatives such as the EU Screening Regulation, the draft of a new further reaching FDI screening legal framework, and the European Economic Security Package,²⁶ which the Compass Strategy also refers to.²⁷ 

The EU FDI legal framework outlines specific factors that should be considered by Member States when determining whether an FDI may pose a risk to security or public order. Among these, the EU rules explicitly mention "critical technologies, [...] including artificial intelligence."²⁸ Additionally, the EU rules refer to critical emerging technologies and critical infrastructure in the range of AI, such as Data Centres. 

Currently, most Member States already require foreign investments in the field of AI to be notified under their FDI rules. However, there is no uniform understanding across Member States regarding the definition of AI and which specific AI-related business activities are subject to notification. While some jurisdictions (such as Germany) have stand-alone tightly defined sectors (only development or production of AI products, capable of being used for cyber-attacks, spread of fake news, mass surveillance or repression), most jurisdictions only refer to the EU Screening Regulation and its undefined sectors. 

The current uncertainties surrounding FDI screening of AI-related investments affect the ability of foreign companies to invest in the EU and scale up the EU’s resilience and independence in this area. To achieve the goals of the Competitive Compass, we expect the EU and Member States will have to establish clearer rules. In particular further clarity could be achieved by the proposal for a new EU Screening Regulation,²⁹ which envisages that all Member States will have to implement a screening mechanism, and establishes a minimum sectoral scope that Member States must apply when assessing foreign investments. This includes critical technologies such as AI, related technologies and infrastructure, as well as projects and programmes of Union interest. 

Export Control and/or Outbound Screening of AI 

The Compass Strategy refers to export controls and outward investment monitoring as tools that could help reduce its dependence on critical input from third countries and promote the EU’s competitiveness. While the current EU legal framework already accounts for the application of export control and outward investment monitoring to strategic technologies such as AI, the EU will need to increase its efforts in this area to keep up with the US, if it wants to deliver on the outcomes identified in the Compass Strategy. 

Regarding export control, the main regulation at the EU level is the Dual-Use Regulation,³⁰ which is not designed for economic protectionism but rather for the control of dual-use goods - i.e. products with both civilian and military applications. While AI can be embedded in some listed dual-use goods, there are no licensing requirements for the export of AI technology in general, although this might be subject to change³¹. However, certain AI-related products are included in the regulation, such as supercomputers exceeding a specific performance threshold. Overall, this regulatory focus highlights the EU’s risk-based approach to export controls, though it may leave gaps in addressing AI’s strategic and economic impact on global competitiveness.

While the US has had an outbound screening regime in place since early 2025,³² the EU has yet to implement such a framework. At this stage, the EU asked Member States to conduct an assessment on the potential establishment of an EU-wide outbound investment screening mechanism.³³ If the EU were to introduce an outbound screening mechanism, and if its final version were to follow a similar approach to the US regime, it could help close the regulatory gap for purely civilian AI applications. Furthermore, an EU outbound regime could make it more difficult for EU investors to finance non-EU entities in certain sectors, such as AI. This, in turn, could incentivise EU investors to focus more on domestic AI development, strengthening Europe's competitive position in the field. However, at this stage, when, whether, and in what form an EU outbound screening regime will materialise remains uncertain. 

Introduction of Europe first requirements in public procurement rules

The Compass Strategy announced a revision of the EU Procurement Directives by 2026, including the introduction of a Europe-first requirement³⁴, in order to reinforce technological security and domestic supply chains, while at the same time simplifying and modernising procurement rules (in particular for start-ups and innovative companies). A more detailed analysis of this reform can be consulted here. Whilst it is still unclear which sectors will benefit from the more preferential regime, the EC has stated in the Compass Strategy that this will cover strategic sectors and technologies. It is therefore likely that, in the coming years, AI projects from EU Companies could benefit from a preferential treatment in public procurement procedures, compared to those of undertakings from third countries.

Rigorous screening of foreign subsidies granted in key sectors, such as AI

The Compass Strategy encourages the EU to continue its rigorous enforcement under the Foreign Subsidies Regulation (“FSR”),³⁵ to target potentially unfair competition from companies benefiting from third country subsidies. Vice President Ribera, responsible for the Clean, Just and Competitive Transition confirmed that the FSR will be one of the top priorities on the agenda of the EC³⁶. 

Companies can expect a strong focus from the EC to review foreign subsidies granted by third countries for AI projects. While the main focus of the FSR will likely remain China, it is expected that FSR will be one of the legal weapons that the EC could rely on to counter the protectionist measures taken by the new US administration, notably with regard to the development of AI.³⁷ This could lead to tougher screening of M&A activities and procurements in the EU by companies that received subsidies from the US, on the basis of the FSR.