Europe's AI Safety Silence Is a Governance Problem
- Europe has been largely absent from the top-level AI safety debate because it lacks homegrown frontier AI companies, leaving it dependent on US and Chinese technology despite its regulatory ambitions.
- The EU AI Act sets requirements for high-risk systems but is not viewed by legal experts as a global benchmark, partly because of its compliance burden and territorial scope.
- Christine Lagarde has described Europe's AI position as a stark choice between forfeiting growth by shunning AI and risking dependency by embracing foreign-developed tools.
- Frederike Kaltheuner of the AI Now Institute argues that American tech executives have framed AI narrowly around frontier models, obscuring the reality that many companies avoid such systems due to cost and data-protection concerns.
Europe has been largely absent from the global conversation about how to keep AI safe. Not because it lacks ambition, but because it lacks the technology companies that set the terms of the debate.
A September 2026 Guardian report on Europe’s sidelined position in AI safety discussions describes a structural problem. Christine Lagarde, head of the European Central Bank, framed the choice starkly: shun AI and forfeit growth, or embrace it and risk dependency on foreign-developed tools. Ursula von der Leyen proposed inviting frontier AI laboratories to discuss development pacing, though such efforts gained little traction without backing from Washington or Beijing. The continent regulates more ambitiously than any other jurisdiction, yet the companies building the most capable systems are American and Chinese. Regulation without the underlying technological base to inform or enforce it is a different kind of power than regulation built on domestic capability.
The EU AI Act sets requirements for high-risk systems and has been presented as a potential global template. But the article notes that legal experts do not view it as a benchmark partly because of its compliance burden and territorial scope. That matters. A law that shapes behaviour within one market but does not influence how systems are designed at the frontier occupies a particular kind of space. It protects consumers locally while leaving the development agenda elsewhere. Frederike Kaltheuner of the AI Now Institute argues that American tech executives have framed AI narrowly around frontier models, obscuring the reality that many companies avoid such systems due to cost and data-protection concerns. The absence of major European frontier labs does not mean the continent has nothing to contribute to the debate about AI’s societal effects. It means the debate’s terms are being set elsewhere.
The dependency is not abstract. European institutions rely on American infrastructure from search engines to chips. Margrethe Vestager, the EU’s former competition commissioner, co-authored a report calling for greater European AI investment, including expanded datacentre capacity, while acknowledging that existential AI threats require global solutions rather than regional responses. The gap between regulatory ambition and technological capacity raises a question about whether governance can be effective when it depends on the very systems it seeks to govern.
The practical consequence for anyone concerned with AI in public life, including education, is that the rules governing how these tools are deployed in classrooms and assessment will be shaped significantly by decisions made outside Europe. Universities developing AI policies may find themselves navigating frameworks built by regulators who had input into the design of the systems themselves. Europe’s challenge runs deeper than competitiveness. It is whether a region can meaningfully govern what it does not build.
The Guardian’s report on Why Europe has been absent from the great AI safety debate provides the source reporting for this article.