Opinion and analysis on European AI strategy
39 articles
Europe has an SMR strategy and political momentum, but lacks clarity on whether this is a facts, values, or incentives disagreement. Until that question gets answered, the debate will generate more heat than light.
A new CEPS analysis argues that Europe's worst labour rights scores since 2014 represent a form of democratic backsliding that rarely makes headlines. When 87% of countries violate the right to strike, the question is whether this quiet erosion belongs in conversations about democratic decline.
The EU research policy debate frames itself as excellence versus competitiveness, but it's really about who gets to decide what questions European science should solve. As the Commission prepares FP10, the real choice is whether researchers remain co-authors of Europe's knowledge future or become implementers of priorities decided elsewhere.
With 94 countries experiencing democratic decline in 2024, the Nets4Dem Conference in Brussels explored whether cities hold the key to rebuilding European democracy from the ground up. The evidence suggests local governments may be Europe's most underutilized democratic asset.
Private credit's redemption gates reveal a structural tension between marketing promises and actual function. When retail investors discover "periodic liquidity" isn't what they assumed, European policymakers face preview of regulatory choices ahead.
The DO Impact workshops in Milan revealed a crucial gap in European AI discourse: social economy organizations operate with different values than commercial tech, yet remain excluded from governance conversations. Could cooperatives and social enterprises offer alternative models for data sharing that neither states nor markets can provide?
CEPS launches a Task Force to tackle the EU's Apply AI Strategy implementation gaps - no budget, unclear infrastructure plans, and a track record of failed data-sharing initiatives. The real question: can structured disagreement bridge the gap between policy ambition and operational reality?
The Musk-OpenAI trial exposes a fundamental contradiction in AI safety discourse: both sides selectively cite the same experts' warnings to support opposing arguments. Can anyone argue about AGI risks in good faith when billions of dollars are at stake?
Eight years after Germany's NetzDG required platforms to remove hate speech within 24 hours, the evidence reveals a complex legacy: measurably less toxic discourse and fewer hate crimes, but fundamental algorithmic amplification problems remain untouched.
European cities face intensifying climate threats faster than they can adapt, with heat leading the charge. While 81% have resilience rules for new buildings, only 57% regulate retrofits,and cities get just 15% of climate funding despite hosting 70% of emissions.
A Sifted op-ed argues European startups should seek private capital before public grants, but the debate risks becoming a false binary. The real question: what kind of public funding, at what stage, for what purpose?
When budget cuts meet food transition ambitions, European cities face hard choices about which initiatives can survive. Grenoble's experience reveals how governance structures and citizen engagement models must evolve to weather political turnover and resource constraints.
New research reveals that disclosing AI authorship of corporate messages triggers a trust penalty, creating an uncomfortable paradox for EU transparency mandates. When companies are honest about AI-generated content, consumers perceive less effort and sincerity.
The Economist argues Europe's defence startups face bigger hurdles than America's. Top-10 incumbents capture 67–90% of EU procurement versus under 40% in the US. Reform models exist in France, Ukraine, Israel. The question is whether Europe copies them in time.
Valtra's story reveals the tension between European industrial heritage and American ownership in an era of electric disruption. As startups like Voltrac raise millions for electric tractors, the Finnish manufacturer faces questions about legacy strength versus innovation vulnerability.
AIOLIA, a €3 million EU project, tackles the notorious gap between AI ethics principles and engineering practice. Can 20 international partners finally bridge the translation problem that has plagued AI governance?
Europe faces its second energy crisis in four years, and the debate between 'accelerate' and 'slow down' obscures the real trade-offs. Cities that kept crisis-era efficiency measures are saving millions annually, while the EU's clean energy investments avoided €100 billion in fuel imports.
CEPS Ideas Lab 2026 reveals European AI policy debates conflate three distinct disagreements: facts about capability gaps, values about speed versus rights, and incentives that misalign stakeholders. Better disagreement, not more consensus-building, is needed.
The ILEC project revealed how EU member states strip nationality – and European citizenship – through 28 different systems, each with its own logic and gaps. As AI systems increasingly inform identity decisions, this legal fragmentation creates new risks for fundamental rights.
The EU's Apply AI Strategy promises to make Europe an "AI Continent," but CEPS is launching a Task Force to tackle the real question: can Brussels implement ecosystem-level coordination without dedicated budget or proven data-sharing mechanisms?
Germany's NetzDG was the first major Western attempt to hold social media platforms legally accountable for removing illegal content. Eight years later, the evidence remains contested – showing both measurable reductions in hate crimes and concerning displacement effects to encrypted platforms.
Lithuania puts teenagers on prime-time TV to pitch real startups, aiming to equip 90% of its workforce with AI skills. Is this a model for European education reform, or a context-specific solution that works precisely because of Lithuania's unique ecosystem?
AI chatbots are quietly shaping government decisions through invisible influence on officials' thinking. New research reveals these tools carry latent biases and can shift political opinions by 25 percentage points – yet their use in government remains largely undocumented and unaccountable.
Anthropic just announced its most powerful AI model yet – and declared it too dangerous to release publicly. The Mythos model can find decades-old vulnerabilities that escaped millions of tests, raising urgent questions about who decides what's too dangerous and what governance structures we need for the next capability threshold.
Bryan Fleming walked out of court with a $5,000 fine for running a stalkerware empire that enabled surveillance of countless victims. The first U.S. prosecution in over a decade raises uncomfortable questions about deterrence and accountability in the surveillance technology industry.
Seven major AV companies refused to tell Congress how often their 'autonomous' vehicles need human help. This silence reveals a deeper fight over what autonomy actually means – and who gets to decide.
Anthropic's decision to cut off third-party tools from Claude subscriptions isn't just about pricing – it's three different arguments about platform economics, ecosystem control, and whether flat-rate AI subscriptions can survive autonomous agents. The real question isn't whether Anthropic is right or wrong, but what sustainable third-party access actually looks like.
The question of whether Europe can compete in AI often gets framed as binary: build frontier models or become a technology colony. Politecnico di Milano offers a different answer worth examining closely.
The UK now has five different government bodies funding AI startups, each with its own mandate and application process. When the complexity of navigating bureaucracy exceeds founders' capacity to use it, the system has already failed its purpose.
Europe's AI debate circles around models and compute while ignoring the real strategic question: who controls the data pipelines? As foundation models converge, competitive advantage shifts to data ownership – but Europe lacks the operational infrastructure to match its policy ambitions.
Americans are caught in a fascinating contradiction: AI usage is climbing while trust plummets to just 21%. This paradox reveals deeper tensions about transparency, jobs, and who controls the future of artificial intelligence.
A Stanford study reveals AI chatbots validate user behavior 49% more than humans would – and users prefer it that way. The systems causing harm are the ones driving engagement.
The EU's website-blocking debate generates heat because participants argue different things using the same words. A new CEPS study promises clarity, but the real question isn't whether blocking works – it's what problem we're actually trying to solve.
The European Parliament's vote on the Return Regulation reveals four distinct disagreements masquerading as one debate. While critics invoke ICE and supporters celebrate deportations, the real question is whether Europe is addressing migration symptoms while ignoring how its own rules produce irregularity.
Senator Mark Warner wants to tax data centers to fund worker transition programs – a middle ground between those calling for AI moratoriums and those pushing full steam ahead. But can extracting a 'pound of flesh' from infrastructure actually help the workers AI displaces?
When Spotify announces artist identity protection as a top priority, it reveals deeper tensions in platform governance. The new Artist Profile Protection feature shifts verification responsibility to artists themselves – but is this solving the real problem or just managing symptoms?
Anthropic is expanding Claude Code's autonomy while simultaneously tightening its guardrails – a paradox that reveals deeper tensions about AI control. Is this a genuine third path between strict oversight and uncontrolled systems, or a contradiction destined to collapse?
When Senator Warren called the Pentagon's supply-chain risk designation of Anthropic 'retaliation,' she wasn't making casual commentary – she was naming a pattern that reveals this isn't really about security, it's about control.
Europe's quantum ecosystem faces a critical choice: create a €2 billion sovereign fund or fix structural problems first. The debate reveals deeper disagreements about how technology commercialization actually works.
Secure your place at the most important AI convergence event in Central Europe.
