The European Union has unveiled three new initiatives aimed at embedding generative artificial intelligence into the daily workings of public administrations across the bloc. With a combined budget of €36.8 million, half of which comes directly from Brussels, the projects — Floods and droughts, Eunomia.AI, and EuropAI — are scheduled to deliver results by 2029 or 2030. The ambition is clear: make European governments more efficient, more responsive, and less dependent on foreign tech giants. But the real test lies not in the funding announcements, but in whether these projects can avoid the pitfalls that have plagued previous EU-backed innovation schemes.
History offers a cautionary tale. In the past, EU funding for digital transformation has sometimes produced little more than glossy reports and pilot projects that never scaled. Critics have coined the term “innovation-washing” to describe initiatives that promise breakthroughs but deliver marginal outcomes. The same risk now applies to artificial intelligence. Without rigorous oversight and the active participation of European companies that actually build AI systems, the new wave of GenAI funding could end up as another exercise in bureaucratic self-congratulation.
Three projects, three approaches
The first project, Floods and droughts, is led by Italy and brings together partners from France, Spain, and Greece. Its €8.8 million budget (€4.4 million from the EU) will fund GenAI tools designed to improve anticipation of climate disasters, crisis management, and communication with citizens. With extreme weather events becoming more frequent across the continent — from wildfires in southern Europe to record-low river levels in Hungary — such tools could prove genuinely useful. But the project’s success will depend on whether the tools are actually adopted by local authorities, not just developed in a lab.
The second initiative, Eunomia.AI, is spearheaded by Spain and involves 13 other EEA countries, including Austria, Estonia, Finland, Germany, the Netherlands, and Norway. With €13.9 million in total funding (€6.9 million from the EU), the project aims to deploy trustworthy GenAI in 25 public administrations and at least 15 public services. It also seeks to bolster European tech sovereignty by relying on open-source models developed within the EU. The infrastructure will be hosted at the University of Granada, with the option to set up federated local nodes. The first pilots are expected within 18 months.
The third project, EuropAI, is led by the Netherlands and includes Denmark, Belgium, and Luxembourg. Its €13.9 million budget (€6.9 million from the EU) will focus on creating a framework for member states to co-create, test, and share GenAI solutions in three areas: legislative simplification, urban planning, and digital citizen assistance. The consortium’s metaphor is telling: they compare Europe’s current reliance on non-EU AI tools like ChatGPT, Copilot, or Gemini to “imported AI fast food” — expensive, opaque, and misaligned with European values. Their alternative is a “community kitchen” where member states cook up their own solutions, using Denmark’s OS2ai platform (already serving 28,000 employees) and the Netherlands’ GovChat-NL as the backbone.
The private sector’s missing seat at the table
While the projects are commendable in their ambition, a glaring issue remains: the limited involvement of European private-sector companies that have real-world experience building and scaling GenAI solutions. In the past, startups and scaleups have often been shut out of EU and national tenders due to Kafkaesque paperwork and bureaucratic hurdles. If the same pattern repeats, these projects risk becoming playgrounds for large consultancies rather than engines of genuine innovation.
The EU’s own regulatory framework, including the AI Act’s deepfake labeling requirements, adds another layer of complexity. While the rules are meant to ensure trust and transparency, they also impose compliance burdens that smaller players may struggle to meet. The challenge is to strike a balance: encouraging adoption without stifling the very innovation the EU hopes to foster.
Moreover, the oversight of where and how EU funding is directed will be crucial. Without clear metrics and accountability, the new projects could easily fall into the trap of “AI-washing” — presenting superficial AI integration as meaningful progress. The EU must ensure that the €36.8 million translates into measurable improvements in public services, not just another set of pilot projects that gather dust.
As the continent faces a warmer-than-normal August and the Danube hitting record lows, the urgency of climate-related AI tools is undeniable. But the EU’s GenAI ambitions will only survive contact with reality if they are grounded in practical outcomes, inclusive of the private sector, and subject to rigorous evaluation. The clock is ticking — 2029 will be here before we know it.
