Views: 191
The EU is responding to the challenges of the digital transformation with an unprecedented number of regulatory acts. Among central objectives of the Union, e.g. the creation of the single market and other integration sub-unions, the EU-wide rules and values also apply to the digital transition’s process. By adopting landmark regulations, the Union wants to strengthen its digital sovereignty and influence in the world, instead of following the digital rules of other states.
Background
In a shared cooperation between the European Commission and EU member states the “EU Coordinated Plan on Artificial Intelligence” has been developed in 2018, aimed to facilitate deployment and use of AI models in a responsible way in the states. Revised in 2021, the plan envisioned mobilising at least € 20 billion in combined annual investments by 2030, to strengthen European global position in trustworthy AI, and to ensure that AI contributed to national sustainable and perspective growth.
For example, new OECD report “Progress in Implementing the European Union Coordinated Plan on Artificial Intelligence (Volume 1): Member States’ Actions”, analyses the national strategies, investments and governance arrangements shaping AI policy in the EU-27 and offers a comprehensive overview of AI’s policy initiatives including data, computing, skills, as well as issues on supporting AI adoption and innovative AI ventures.
On the EU coordinated plan in: https://www.oecd.org/en/publications/2025/11/progress-in-implementing-the-european-union-coordinated-plan-on-artificial-intelligence-volume-1_bb04a5b6.html
Full OECD report in: https://www.oecd.org/en/publications/progress-in-implementing-the-european-union-coordinated-plan-on-artificial-intelligence-volume-1_533c355d-en/full-report.html
Almost all EU states (24 out pf 27) have already adopted national AI strategies; the remaining three are on the way to “join the club); however, still less than half the EU states have specific budgets for implementing AI strategies: generally, the AI models are funded through broader digitalisation programs or sector-specific measures. The lack of consistent “digital budgets” makes it difficult to manage the AI-specific expenditures and monitor progress.
OECD experts note that AI’s effectiveness depends heavily on data quality, availability and computing capacity: as soon as the EU states have adopted national data strategies, both the recognition of big data coverage as a “public good” and an economic asset is becoming valid; many states promote open data policies, secure data-sharing frameworks and interoperability standards to support the AI development.
Meanwhile, the demand for secure, high-performance digital infrastructure is growing fast: i.e. over two-thirds of EU states are strengthening high-performance computing (HPC) capabilities, often through the EuroHPC Joint Undertaking, and investing in edge computing, sovereign cloud models and connectivity networks to support AI workloads. More than half EU states support semiconductor R&D and manufacturing, in line with the objectives of the European Chips Act. These initiatives aim to ensure the EU-wide digital sovereignty and global competitiveness in AIs. The EU states’ sectoral priorities are generally aligned with those identified in the EU Coordinated Plan: i.e. healthcare, public administration, mobility, agriculture, climate and environment; however, the AI intensity are varied across national priorities.
Source: https://oecd.ai/en/wonk/tracking-europes-progress-on-ai-insights-from-the-implementation-of-the-eu-coordinated-plan-on-artificial-intelligence
EU’s big digital acts
There are at least eight so-called “big digital acts” in the European digital regulatory framework (though, generally, there are over 50 EU-wide “digital rules”):
1. Data Governance Act (2023), 2. Cyber and Data Protection Act (2023), 3. Digital Market Act (DMA) and 4. Digital Services Act (DSA), 5. Data Act (2023), 6. Digital Company Act (upgraded), 7. Digital Company Law Directive (to be transposed to the member states by July 2027), and, finally, 8. European Artificial Intelligence Act (EU AI Act, 2024 in 113 articles, delayed until December 2027, coped with the AI Liability Directive).
There is as well the so-called “Common European Data Spaces” initiative to be implemented by 2027): i.e. the EU aims to create the continental-common, interoperable data spaces in strategic sectors to overcome existing legal and technical barriers to data sharing. The rules for common European data spaces will cover areas like health, mobility, environment, energy and agriculture, to: a) make better use of publicly held data for research for the common good, b) support voluntary data sharing by individuals, and c) set up structures to enable key organisations to share data.
More in: https://www.freshfields.com/en/our-thinking/campaigns/tech-data-and-ai-the-digital-frontier/eu-digital-strategy/european-data-spaces
The EU adopted already in July 2022 two landmark digital regulations, the DMA and the DSA; both pieces of legislation codify principles and create processes for digital services providers to follow and ensure a wide choice of safe digital services and derivative products.
More on DMA/DSA in: https://eizpublishing.ch/artikel/euz/03-2024/digital-regulation-in-the-european-union/
The European AI Act (EU AI Act)
The EU is positioning itself as a global leader in trustworthy, human-centered artificial intelligence. The EU’s regulatory framework aims to balance innovation with safety, transparency, and protection of fundamental rights. Europe’s AI regulatory framework aims to create safe, ethical, and trustworthy AI while supporting innovation. The EU AI Act is becoming a global reference model, influencing regulatory discussions worldwide. European Union aims to lead in trustworthy, human-centered AI through strong regulations (often excessive); the ambition is clear: to balance innovation with citizens’ safety, data protection and fundamental rights. But — as with any ambitious regulation — there are debates, trade-offs and dissenting voices.
Reference -the EU AI Act (published, June 2024, Regulation 2024/1689): https://ai-act-law.eu/
The risks
The AI Act applies a risk-based approach to digital developers: e.g. stricter regulation for higher-risk AI systems, and lighter or no rules for lower-risk AI systems.
= Unacceptable risks (prohibited and banned in the EU states): “social scoring” by national governments, including, e.g. manipulative or exploitative AIs, certain biometric surveillance practices.
= High risks: the AI models that actually/potentially affect safety and/or fundamental rights, including: critical infrastructure, medical devices and services, employment/recruitment systems, law enforcement offices/tools. Some requirements include: strict data quality rules, human oversight, robust documentation, as well as cybersecurity and risk management.
= Limited risks include chatbots and AI systems that interact with humans; the main requirement is transparency, i.e. the customers/users must know they are interacting with AI.
= Minimal Risk, including most AI applications (e.g., spam filters, AI games, etc.); there is no regulatory obligations, although some voluntary codes of conduct are encouraged.
The EU AI Office
As a newly created EU body/office, it is supervising high-impact and general-purpose AI (GPAI) models. Besides, it cooperates with national authorities to ensure consistent enforcement; it issues guidance, technical specifications and performs evaluations of major AI models.
According to art. 112, by 2 August 2028, the Commission shall evaluate the functioning of the AI Office, whether the AI Office has been given sufficient powers and competences to fulfil its tasks, and whether it would be relevant and needed for the proper implementation and enforcement of the AI law, to upgrade the AI Office and its enforcement competences and to increase its resources.
Obligations for General-Purpose AI Models
Transparency requirements on training data; reporting capabilities and risk assessments; additional obligations for “systemic” GPAI (large-scale or high-impact models). According to art.51, a GPAI model shall be classified as a “general-purpose AI model with systemic risk if it meets any of the following conditions: a) it has high impact capabilities evaluated on the basis of appropriate technical tools and methodologies, including indicators and benchmarks; and b) if it has capabilities or an impact equivalent to those set out in Annex XIII”.
For example, in the UK law, there are seven GDPR principles: Lawfulness, fairness and transparency; Purpose limitation; Data minimization; Accuracy; Storage limitation; Integrity and confidentiality (security); and Accountability.
Reference to: https://ico.org.uk/for-organisations/uk-gdpr-guidance-and-resources/data-protection-principles/a-guide-to-the-data-protection-principles/
Enforcement and penalties (AI Law, chapter 12)
The EU member states shall lay down the rules on penalties and other enforcement measures, which may also include warnings and non-monetary measures, applicable to infringements of the law by operators, and shall take all measures necessary to ensure that they are properly and effectively implemented, thereby taking into account the guidelines issued by the Commission pursuant to art. 96. The penalties provided for shall be effective, proportionate and dissuasive; they shall take into account the interests of SMEs, including start-ups, and their economic viability.
Thus, fines are scaled to company size, similar to GDPR; penalties can reach: up to €35 million or 7% of global turnover for the most serious violations. Thus, there are strong incentive for compliance across the AI supply chain.
Some controversies
There are already influential voices (e.g. from former EU leaders, industry experts, etc.) calling to “pause” certain AI Act’s provisions or delay its full implementation, citing uncertainties, investment concerns and data-access constraints.
On the other hand, many civil-rights and consumer-protection organizations warn that delaying or weakening the rules would undermine fundamental protections: i.e. arguing that “simplification” should not be at the expense of human rights and general oversight.
The public-private “tensions” reflect a core dilemma: how to regulate AI so as to protect citizens and human/corporate rights without crippling innovation or making European states uncompetitive. For example, the so-called “30 percent rule” in any research work means that only 30% of the content shall be from the AI tools and 70% – from learners’ own ideas and efforts.
More in: https://www.integrin.dk/2025/10/26/european-ai-act-controversies-problems-in-investments-and-personal-data/
Balanced approach
As soon as the AI Act represents a landmark and ambitious attempt to govern AI responsibly, it is so-to-say laudable and globally pioneering. But regulatory ambition comes with costs and trade-offs: high compliance burdens, potential barriers to innovation, risk of centralization, and possible unintended chilling effects on smaller actors and open-source development.
The success of the regulation will depend heavily on how it is implemented: clarity of definitions, fairness of enforcement, support mechanisms for SMEs/startups, and accommodating diversity in AI development (not just centralized big-tech models).
The current debate around postponements or reform proposals shows that the regulatory balance is still being negotiated — and that the final shape of Europe’s AI ecosystem remains uncertain.
More in: https://artificialintelligenceact.eu/ai-act-explorer/
Our opinion
Europe seeks to lead with safe, ethical and transparent AI; but this ambition must grapple with real economic, technical and social trade-offs. The AI Act is a strong foundation, although this regulation is not self-executing: its impact will depend on enforcement, clarity and ongoing dialogue among regulators, industry, civil society and a wider public.
Recognizing and engaging with critiques is vital for shaping a durable, balanced AI ecosystem in Europe that protects rights, encourages innovation and remains globally competitive.
It seems that ST. Hawking were right warning that “if AI evolves beyond human control, it could fundamentally change or even endanger the future of humanity”; in the EU, Italy was the first to ban DeepSeek’s AI.
More in: https://artificialintelligenceact.eu/