A contribution from Oskar J. Gstrein
A. Introduction
Artificial Intelligence (AI) is everywhere. In work and finance, AI promises to increase efficiency and make organizations more productive,[1] while raising concerns such as graduates not getting entry-level jobs to develop their careers,[2] lawyers being replaced by large language models,[3] humanoid robots ousting workers in car factories,[4] or financial markets crashing once the AI bubble bursts.[5] In education, tailored AI tutors pave the way to more engaging and productive personalised learning,[6] whereas in ‘AI powered schools’ entire curricula are drawn up automatically,[7] with students and lecturers duelling each other in leveraging their skills prompting chatbots to create and mark written assignments. In research, AI enables the finding and development of new drugs,[8] amongst many other things. In warfare, lethal autonomous weapons systems automate threatening, harming, and even killing people, raising questions about meaningful human control.[9] And for those who find themselves confused, in agony, and craving for emotional affection there is ‘artificial intimacy’[10] – coming in all shapes and sizes, capable of endless attention, just waiting for its users to share their feelings, being taken on dates, flirt with, etc.
It was in 2017 when the heads of states of EU countries first considered AI governance and regulation as a subject worth pursuing,[11] eventually leading to the creation of an AI High Level Expert Group[12] which famously published ethical principles[13] and investment guidelines[14] in the subsequent years. The proactive stance of policy makers in making governance an integral part of AI design was promising at first, seemingly countering the narrative of a regulator that is always lagging in the digital domain. But in early 2026, it seems that all this carefully prepared groundwork for an AI law – a 2024 EU Regulation called the AI Act[15] in the spirit of the ‘act-ification’[16] of Union Law – has stalled with the presentation of the Digital Omnibus on AI Regulation Proposal on 19 November 2025.[17]
As the legislative wheels in the European Parliament and Council slowly begin to turn,[18] this contribution presents a few highlights of this remarkable journey throughout which the global powerhouse of technology regulation – the EU legislator with the allure of the ‘Brussels effect’[19] – has lost its way in AI Regulation. It argues that this unpredictable turn in technology regulation significantly undermines the status of the bloc as a champion of technology regulation on a democratic basis, while actors shaping and deploying AI move swiftly to establish facts, and ask questions later.
B. Chapter I: The Grande Ambition
With the presentation of the EU AI Act proposal in 2021,[20] Europeans were promised safe and trustworthy AI systems, which are transparent, resilient, and human-centric.[21] Maybe that would mean that the latest trends relating to AI would no longer emerge in Europe, but at least European citizens were guaranteed that they would only encounter ‘mature’ AI in their lives (why would you want an AI-enabled toy that spies on your children, anyways?).[22] However, in contrast to the fundamental-rights-focused 2016 EU General Data Protection Regulation, very soon it became questionable[23] whether the actual regulatory objective of the AI Act would be fostering fundamental rights and human dignity,[24] or merely becoming a technical standardisation framework facilitating the internal market following Article 114 TFEU. Today, it seems established in the literature that the 2024 EU AI Act is a ‘hybrid regulatory framework’ combining product standardisation with the protection of fundamental rights (in particular Article 16 TFEU).[25]
While not perfect, such a hybrid nature with a risk-based approach[26] (complicated to apply for general-purpose AI[27]) could still have been an acceptable – pragmatic – take on AI Regulation. After all, the strength of the EU lies in the regulation of markets and fostering economic integration. In addition, the reliance on mostly existing and known principles such as standardisation – which applies to a wide range of the ‘systems’ which are part of the high-risk category anyways – might also be seen as an attempt to push back on AI exceptionalism and treat the technology more similarly to others which are part of our everyday lives.[28]Unfortunately, as global tensions arise,[29] with European businesses concerned about the economic outlook,[30] and overburdened public authorities incapable of delivering implementation guidelines and standards along the originally proposed deadlines,[31] that grande ambition has died.
C. Chapter II: Death by a Thousand paper cuts
What is the motivation behind the Digital Omnibus on AI? It is part of a broader EU Commission initiative with ten ‘packages’[32] (with the toughest implications for the sustainability sector) to review legislation to improve European competitiveness by saving costs for businesses and boosting innovation.[33] It is clear how reducing obligations for compliance merely consisting of administrative work – rather than re-shaping and structuring how data and technology is being used more safely and efficiently – can immediately reduce costs for business. But reducing safeguards and rules guiding the design of AI systems does not directly lead to more innovation. Rather, there is a lack of investment and infrastructure to have AI not only researched, but also deployed in the EU by European actors.[34]
The most worrisome aspect about the Digital Omnibus on AI is its piecemeal approach, further compromising the vision of AI regulation in Europe. Looking at the highly informative ‘staff working document accompanying the proposal’,[35] the methodology of the EU Commission becomes clear. On the one hand, there is a high-flying political narrative that there needs to be less regulation. On the other hand, specific pain points for stakeholders in the 2024 AI Act were identified throughout workshops. This forms the basis for the specific amendments which have been tabled, in the form of very technical changes to a not fully implemented law which are difficult to understand and interpret, even for scholars studying this field for years. The problem – or, more positively formulated, the challenge – for the European Parliament and the Council is now to bring back a political narrative to the story, based on democratic representation and constitutional principles.
There might be many reasons why the EU lost its way in regulation relating to data and technology, including AI FOMO,[36] a wrongly understood approach to ‘simplification’, or an attempt to attack fundamental rights and especially data protection.[37] Regardless of which of these reasons prevails – or whether it is a combination of all of them – it is at this point completely unclear what the story of the AI Act is, and for whom it matters. The story has lost its plot. Businesses are unhappy,[38] civil society is unhappy,[39] governments are unhappy.[40]
D. Chapter III: Back to square one
It is a well-known cliché that AI Regulation is mainly about the question whether it stifles or fosters innovation. But this is an overly simplified view on technology regulation, and the impact AI already has on society. As mentioned at the very beginning, AI is everywhere. As the benefits and the risks of this technology affect all our lives, we all should participate in how that happens – right from square one. That exactly was the promise of having a democratically legitimized and horizontally applicable law on AI; for once a legislator that is not running behind developments (ex-post), but a proactive, constitutionally and democratically legitimised, and co-creational process determining how the design of socio-technical systems empowers citizens (ex-ante).
Unfortunately, the tabling of the Digital Omnibus on AI Regulation Proposal with its lack of vision, thereby further watering down what already had become a ‘hybrid regime’, and combined with an overly responsive and defensive stance of the EU Commission, has undermined this noble objective. The Digital Omnibus on AI administers an overdose of legal uncertainty, while exposing the weakness and indecisiveness of the EU legislator in an area of not only technological, but strategic importance.
AI regulation in the EU was never going to be perfect, but that was not necessary. Having a constitutionally legitimised legal framework based on democratic governance would give regulation a meaningful stake in shaping how AI develops, ultimately empowering the people who legitimize the constitutional process to have actual influence in how these systems shape their lives. But now, with the publication of many technical changes lacking an overall vision, and further delay in a process that could go in many different directions, the ball is back in the court of those who ‘move fast and break things.’
E. Final Chapter: Unwritten
It remains to be seen what the next months will bring for AI regulation in the EU. What has already happened is that the indecisive stance of the EU allows countries such as China[41] and the United States[42] to shape this space, with global implications. The EU AI Act still needs to be fully implemented by August 2026, as long as the digital omnibus has not been finally negotiated and passed the Parliament and Council before that deadline. It would be a legislative miracle if this happens in a couple of weeks, which is what businesses and governmental organisations need to start preparing for what to comply with after the summer. For now, only uncertainty remains – and the hope that the legislator will have the ability to not only focus on a reactive stance responding to short-term interests of the loudest voices, but develop a regulatory vision that responds to the transformative challenge that AI presents for democracies and their constitutional orders.
Acknowledgement: This post has been written with support of the NWA DECIDE project (https://decideproject.nl), funded by the Dutch Research Council (NWO) under the Dutch Research Agenda (NWA), project number NWA.1766.24.031
[1] The Decoder, KI Produktivitätsschub in den USA angeblich erstmals messbar, 16.02.2026.
[2] Anjili Raval, The great graduate job drought, 21.01.2026.
[3] Julia Kollewe, Anthropic’s launch of AI legal tool hits shares in European data companies, 03.02.2026.
[4] Osmond Chia, Car giant Hyundai to use human-like robots in factories, 06.01.2026.
[6] Lieke van den Krommenacker, A chatbot as a tutor, 08.02.2026.
[7] Emanuel Maiberg, ‘Students Are Being Treated Like Guinea Pigs‘ Inside an AI-Powered Private School, 17.02.2026.
[8] Dharmasivam, M., Kaya, B., Akinware, A., Azad, M. G., & Richardson, D. R. (2025). Leading AI-driven drug discovery platforms: 2025 landscape and global outlook. Pharmacological Reviews, 100102.
[9] Taís Fernanda Blauth, Autonomous Weapons Systems in warfare is: Meaningful Human Control enough?, in: Zwitter/Gstrein (eds.), Handbook on the Politics and Governance of Big Data and Artificial Intelligence.
[10] Wikipedia, Artificial intimacy.
[11] European Commission, Communication from the Commission to the European Parliament, the Council, the European Economic and Social Committee and the Committee of the Regions on the Mid-Term Review on the implementation of the Digital Single Market Strategy A Connected Digital Single Market for All, COM/2017/0228 final.
[12] European Commission, High-level expert group on artificial intelligence.
[13] European Commission, Ethics guidelines for trustworthy AI.
[14] European Commission, Policy and investment recommendations for trustworthy Artificial Intelligence.
[15] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act), OJ L 2024/1689.
[16] Vagelis Papakonstantinou, The act-ification of EU law: The (long-overdue) move towards eponymous EU legislation.
[17] European Commission, Digital Omnibus on AI Regulation Proposal.
[18] Luca Bertuzzi, EU countries seeking position on AI Act amendments by April, 12.01.2026.
[19] Anu Bradford, The Brussels Effect: How the European Union Rules the World.
[20] Proposal for a Regulation of the European Parliament and of the Council laying down harmonised Rules on Artificial Intelligence (Artificial Intelligence Act) and amending certain Union Legislative Acts, COM/2021/206 final.
[21] European Commission, AI Act.
[22] Andy Greenberg, An AI Toy Exposed 50,000 Logs of Its Chats With Kids to Anyone With a Gmail Account, 29.01.2026.
[23] Michael Veale/Frederik Zuiderveen Borgesius, Demystifying the Draft EU Artificial Intelligence Act, Computer Law Review International (4):2021.
[24] European Commission, AI Act.
[25] Marco Almada/Nicolas Petit, The EU AI Act: Between the rock of product safety and the hard place of fundamental rights, Common Market Law Review 62(1):2025.
[26] Dariusz Szostek/ Rafał Tomasz Prabucki, High-Risk AI Systems, in: Vera Lúcia Raposo (ed.), The European Artificial Intelligence Act.
[27] Oskar J. Gstrein/Noman Haleem/Andrej Zwitter, General-purpose AI regulation and the European Union AI Act, Internet Policy Review 13(3):2024.
[28] Xiaoyao Han/Oskar J. Gstrein/ Vasilios Andrikopoulos/Ronald Stolk, Every wave carries a sense of déjà vu: revisiting the computerization movement perspective to understand the recent push towards artificial intelligence, AI&Society 40:2025.
[29] Jennifer Rankin, EU could water down AI Act amid pressure from Trump and big tech, 07.11.2025.
[30] Barbara Moens/Tim Bradshaw, European CEOs urge Brussels to halt landmark AI Act, 03.07.2025.
[31] EU-Insider, EU considers delaying parts of AI Act as industry awaits guidance, 06.06.2025.
[32] European Commission, Simplification.
[33] European Commission, Simpler EU digital rules and new digital wallets to save billions for businesses and boost innovation, 19.11.2025.
[34] Owen-Carpenter-Zehe, Why is the EU struggling to scale artificial intelligence, 13.02.2026.
[35] European Commission, Commission Staff Working Document, SWD(2025) 836 final, p. 68 onwards.
[36] Felix Bieker/Katherine Nolan, European AI FOMO, 29.01.2026.
[37] Hannah Ruschemeier, The Omnibus Package oft he EU Commission, 17.11.2025.
[38] Digital Europe, AI Act delay is not enough: the omnibus must fix Europe’s industrial competitiveness, 16.02.2026.
[39] Open Letter, The EU must uphold hard-won protections for digital human rights, 13.11.2025.
[40] Luca Bertuzzi, France seeks to split AI Act delay from changes in bid for deeper reform, 23.01.2026.
[41] Nature, China is leading the world on AI governance: other countries must engage, 10.12.2025.
[42] The White House, Ensuring a National Policy Framework for Artificial Intelligence, 11.12.2025.
Suggested Citation: Gstrein, Oskar J., The EU AI Act: A Story without a Plot? A Lack of Vision in the Digital AI Omnibus Proposal threatens constitutional AI Governance, jean-monnet-saar 2026.
DOI: 10.17176/20260324-095903-0
Funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – Project No.: 525576645
