European AI policy today resembles the early days of Covid. The trend lines show we are rapidly moving up an exponential, and with the announcement of Anthropic's Claude Mythos Preview, European leaders are slowly starting to realise it. AI already writes most of the software at the leading AI labs, is beginning to run its own research, and is redefining cybersecurity. Before long, it will disrupt labour markets, warfare, and the global balance of power. The last time a technology changed life as completely as AI is about to, we called it an Industrial Revolution.
Most of Europe has not yet absorbed the pace and magnitude of the change that is coming, and those who have are not saying so loudly enough. It is more comfortable to hope that AI will somehow blow over, that its risks will fail to materialise, or that being clever will substitute for the hard trade-offs needed to stay relevant in a rapidly changing world. But this is not a moment for ordinary politics.
Europe 2031 is an attempt to explain what Silicon Valley is seeing and Europe is not. But it is also an attempt to shake the continent into action, so that we can maintain a say over our own future.
Our story begins in January 2025, with the public release of DeepSeek's R1 model, and runs until March 2031 - by which point Europe has all but lost the ability to chart its own path. It closes with an epilogue set in 2034, looking back on the choices that led to Europe's slide into irrelevance, and on the alternatives still available, today, in June 2026.
The storyline we lay out is not meant to be a prediction. But we have tried to make it internally consistent, technically sound, and traceable to dynamics visible today. The exact dates and events are not the point; what matters is that the kind of future we describe is plausible enough to be taken very seriously.
It is one thing to read about AI's potential impact in the abstract, but it is another to feel it in your bones. That is why we have written Europe 2031 as a novella-like story, told through the eyes of two fictional characters. We have kept the events they experience as realistic as possible, however, and we do not mean to hide from criticism behind a cover of fiction. We almost certainly got things wrong, and we truly welcome substantive feedback. If the story sparks a wider and more urgent debate about transformative AI and Europe's role in the transition, we will have succeeded at much of what we set out to do.
Europe 2031 was written by a small group of AI researchers, think-tankers, and investors who have spent their careers at the intersection of frontier AI and European policy. Among us, we have advised political leaders, helped shape European AI legislation, co-authored national AI strategies for Germany and the Netherlands, built frontier AI models, invested in leading European AI companies, and contributed to the core writing team of the International AI Safety Report.
All contributors participated in a personal capacity, and the views expressed do not necessarily reflect those of their current or past employers or affiliations.
We would like to thank a number of people without whom Europe 2031 could not have happened. None of them necessarily agree with the events and recommendations we describe, and any mistakes are our own.
First, Tom Chivers, whose writing turned a dry scenario into an engaging story that is a joy to read, and who recorded the audio version of the story.
Second, Daniel Kokotajlo, Thomas Larsen, and the rest of the team at the AI Futures Project. AI 2027 was a major inspiration for our piece, and we hope we have done justice to the format it introduced (Europe 2031 is not affiliated with AI 2027 in any way).
Third, Anton Leicht, whose strategic input heavily shaped the final storyline.
Fourth, Chloé Touzet and Bálint Pataki, whose political and institutional knowledge made the scenario significantly more realistic.
Fifth, Katy Sexton, whose support on communications and media outreach was instrumental in helping Europe 2031 reach the audiences we hoped it would.
We would further like to thank the following people for their help and feedback:
Ben Harack, Daniel Privitera, Harry Law, Henry Papadatos, Jaime Sevilla, Jan-Willem van Putten, Oscar Delaney, Philip Trippenbach, Pieter Garicano, Sander Volten, Charles Martinet, and Tim Rocktäschel.
We have written this story because we care deeply about Europe, and because we believe a strong Europe is important for a better world. We have no interest in hype, nor in cynicism. Plenty of smart, well-informed people in Silicon Valley have already written the continent off, convinced its decline is settled and irreversible. Some of our colleagues in the US ask why we don't relocate to California while we still can. We refuse to accept this fate. Though their arguments are persuasive, we think they are too quick to dismiss Europe's strengths. We believe that Europe can still steer this technology, rather than be steered by it — but only if it finds the will to. We hope this scenario helps to garner that will.
The clearest signals about what's next in AI come from a dense network of Silicon Valley elites — circles that Europeans often resent and choose to ignore. San Francisco is home to many smart, well-meaning AI experts. But it is also home to a few Big Tech CEOs who have taken morally repugnant actions, made factually incorrect claims in the past, and who profit handsomely from the AI boom. None of that, however, makes them wrong about AI — in fact, their predictions have aged far better than those of European AI sceptics. We think it is vital to keep listening to the people closest to the technology, despite their conflicted interests. We should scrutinise their claims, but not dismiss them out of hand. Just because AI companies say it's sunny doesn't mean it's dark outside.
We realise our scenario focuses almost exclusively on European sovereignty, and only briefly touches on many other aspects of the AI transition — some of which are arguably more pressing. This was deliberate. We believe Europe must preserve its agency and leverage if it is to do anything at all, which includes mitigating AI risks.
We embarked on this project partly because we think a more empowered Europe will, on the current margin, lead to a safer and better AI transition. Europe has a strong track record as a defender of fundamental rights and values, including in AI. We strongly support the AI Act's General-Purpose AI Code of Practice, and the EU's expert-led process that led to it. We believe a strong Europe can act as a more responsible and positive-sum counterbalance to major powers, while a weak Europe can't.
We have spoken candidly about AI security and safety before. To avoid any misinterpretation, let us be very clear: we believe there is a serious chance that advanced AI goes badly wrong, whether through cyber- or biological misuse, extreme concentration of power, or a failure to align AI systems with human values. In the worst cases, this could lead to human extinction or its permanent disempowerment.
These risks are widely discussed, including in authoritative works like the International AI Safety Report. Our story is by no means an attempt to dismiss them. Rather, we hope to empower Europe so that it can, among other things, mitigate these risks from a position of strength.
Most of the work on this scenario was done by people in their free time. Arq Foundation — where co-authors Daan Juijn and Alex Petropoulos work — provided limited funding for things like a website and professional translation. We received no funding from the AI industry, and we have no profit motive. The author group deliberately spans think-tankers, academics, technical researchers and investors, to bring together a broad range of perspectives.
Restoring Europe's agency will be a herculean effort. Although many are already working hard on this, far more voices are needed: from thought leaders and community builders, to research ideation and agenda-setting, to on-the-ground political operators and funders to accelerate all of the above. To kick-start this field, we need talent, capital and ideas. If you have one of those, but are missing the others — please get in touch at info@europe2031.ai, and we will try and match you with collaborators and philanthropic funders interested in this space. We are already in touch with several.
If you've spotted something we missed, want to talk to one of the authors, or have a question that's not answered above — drop us a line below, or at info@europe2031.ai.