The EU’s bureaucracy has a new obsession: containing the AI wildfire. In May 2024, it introduced the Artificial Intelligence Act, the world's first comprehensive AI regulation. Its aim? To prevent infringing on civil liberties while improving upon security.
Yet, the clash between EU unity and national sovereignty persists. Take the AI Act’s national security exemption, which allows member states to sidestep rules when AI serves law enforcement or defence needs.
Unlike antitrust rules, where Brussels wields real clout, AI oversight splinters across borders. Each country picks its own regulators to monitor AI use in policing, leaving the European Commission powerless to enforce directly. This patchwork breeds uneven standards—AI tools have different controls put on them depending on where they’re deployed.
Different Countries, Different Priorities
France has been at the forefront of AI-driven security for years. Ahead of the 2024 Olympic Games, French authorities expanded the use of AI-powered surveillance, including facial recognition and behaviour analysis. While officials defended these measures as necessary for public safety, civil rights groups warned of excessive state control. The debate reflects broader concerns about how AI may shift the balance between security and individual freedoms.
Germany has taken a more cautious approach so far. AI-driven law enforcement tools are deployed under strict oversight with a strong emphasis on data protection. Efforts to counter online hate speech using AI have accelerated, particularly in response to what German authorities call digital extremism.
Hungary has integrated AI into its border security infrastructure. Automated surveillance systems, combined with AI-driven risk assessments, have been deployed along its southern border. Budapest presents these measures as necessary to safeguard national security, while critics raise human rights concerns.
The Stark Reality of National Sovereignty
Despite its regulatory ambitions, the EU faces structural barriers to AI centralisation. Overall, national governments are unwilling to cede control over policing, seeing it as a core element of their sovereignty. While groundbreaking, the AI Act remains subject to national carve-outs and varying enforcement mechanisms, weakening its effectiveness as a unified framework.
This tension is part of a broader struggle between sovereigntists and federalists, one of the defining fault lines in the EU’s political landscape. Sovereignists argue that deeper integration erodes national decision-making, while federalists push for stronger central governance to enhance efficiency and unity. This divide was a key factor in Brexit, as many in the UK rejected Brussels’ increasing influence over domestic affairs.
Centralisation would mean stripping member states of control over critical areas, including law enforcement, immigration policy, and judicial oversight, all of which are deeply tied to each country’s legal and political traditions. The push for an EU-wide AI policing system is just another example of how this friction plays out, with some states resisting what they see as a Brussels-driven encroachment on national sovereignty.
Legal and structural barriers to make the “dream” come true
The road to AI centralisation is blocked by other obstacles. One major challenge is the varying levels of technological readiness across member states. While countries like France and Germany have invested heavily in AI-driven policing tools, others lag behind due to financial constraints.
This digital divide complicates efforts to create a harmonised AI enforcement system. The European Commission has sought to address this disparity by allocating funding through initiatives such as Horizon Europe, but uptake remains uneven, with some governments prioritising other security concerns over AI development.
One thing is however certain; a major push for AI development across the EU is happening. The European Court of Auditors reports that between 2018 and 2020, the European Commission allocated €500 million per year for AI research and innovation through its Horizon Europe and Digital Europe programmes. From 2021 onwards, this funding doubled to €1 billion per year. While these funds target AI development across sectors, its application to law enforcement remains one of the most hotly contested.
Yet the EU's General Data Protection Regulation (GDPR) places strict limits on data collection and automated decision-making, directly affecting AI applications in law enforcement. Facial recognition, for example, remains highly restricted in many EU countries due to privacy concerns, with some states banning its real-time use in public spaces. These legal barriers mean that even if AI tools are developed and deployed, their effectiveness varies depending on national interpretations of EU privacy laws.
The dream of a single, AI-powered policing model for the EU remains just that—a dream. Brussels may push for unity, but security is still a language composed of many regional dialects. The result isn’t seamless integration, but a patchwork of AI enforcement—shaped less by European ideals, than by national realities.
Statement
AI is transforming law enforcement, but Europe’s legal patchwork is putting Brussels’ ambitions for a centralised system to the test. As national priorities undermine the EU’s push for uniform AI policing, it is questionable whether AI can ever be truly centralised in Europe. As AI policing grows more powerful, this debate will define the balance between security, sovereignty, and civil liberties for years to come.