The European Green Deal runs on data, from corporate disclosures to climate-risk analyses and decarbonization pathways. But much of this data feeds platforms and tools developed outside the European Union, which could threaten the bloc’s climate agenda and even its foundational values.
PARIS—Last October, Signe Ratso, Deputy Director-General of the Directorate-General for Research and Innovation of the European Commission, issued a stark warning while speaking on a panel at the European Parliament. “When critical datasets and analytical tools are hosted outside” the European Union, she explained, “their access or even the content … can change beyond our control.” As a result, “core scientific values can be compromised.”
Ratso’s remarks highlight one of the main issues facing European climate researchers. While many people are focused on the implications of the “backlash” against sustainability for the European Green Deal, these conversations—as important as they are—overlook the less visible, but arguably more significant, problem of data extraterritoriality.
The threat is subtle. The United States has enacted laws that give it effective control over data and digital infrastructure, regardless of where they are located. This suggests a creeping expansion of the US International Traffic in Arms Regulations, which govern the export of military technologies and technical data, into the civilian realm. Such instruments allow the US to exert pressure by imposing traceability and end-use monitoring on sensitive fields like climate research and digital systems. Instead of instituting outright bans, US policymakers can quietly shape what is studied and how.
The EU seems well-equipped to counter this threat. The General Data Protection Regulation, the Data Act, the AI Act, and soon the Cloud and AI Development Act are meant to defend the bloc’s digital sovereignty. But Europe’s regulatory fortress has a foundational weakness: overreliance on foreign-controlled infrastructure and datasets. Equally worrying are Europe’s excessive focus on compliance, rather than operational implementation, and its insufficient real-world testing of usage governance, particularly for AI.
Over the past decade, for example, the share of ESG (environmental, social, governance) and climate datasets and models owned or controlled by non-EU actors has steadily grown. Now, according to the PARC Foundation’s ESG Data Cartography, more than 80% of the ESG data volume used in the EU falls into this category.
This concentration of control creates an asymmetry. Europe writes the rules, including the Corporate Sustainability Reporting Directive and the Sustainable Finance Disclosure Regulation, that are responsible for generating vast amounts of data: corporate disclosures, green taxonomies, climate-risk analyses, decarbonization pathways, and just-transition policies. But the analytical tools that turn these data into power are overwhelmingly non-European.
What makes the asymmetry particularly troubling is its cumulative effect. The raw data feed models that improve with scale and the most advanced systems become ever more dominant and difficult to replace. Over time, this dynamic could lock Europe into a dependency on the foreign-owned infrastructure that captures the economic value of ESG data. Such ownership concentration is not only a sovereignty risk, but also a competitiveness trap.
AI is accelerating the process. While traceability of training data—along with documentation and auditability—is a prerequisite for AI access to EU markets, that is not necessarily how it works in practice, especially because compliance deadlines are still far away. And while it is conceivable that large language models could act as effective safeguards against climate-skeptic narratives on social media, they have tended to exhibit regressive complacency biases and, in some cases, regurgitate falsehoods. Research by PARC shows a significant gap between Grok and DeepSeek in terms of spreading climate misinformation.
Efforts to make these data publicly accessible have fallen short of their potential. The Net-Zero Data Public Utility (now called the Climate Data Utility), launched in 2022 to create a centralized, open repository for climate transition-related data from companies worldwide, has stalled. Likewise, the scope of the European Single Access Point, which is set to go live in 2027, has narrowed amid a simplification push.
Europe’s green transition rests on an informational infrastructure that it does not control. In February, US Energy Secretary Chris Wright reportedly asked the International Energy Agency to abandon its climate scenarios, long regarded as the global benchmark for decarbonization pathways, and the closest thing to a shared net-zero scripture that the world has. This episode highlights how contested an agreed theory of change can become. At stake is more than the execution of specific policies; the EU’s entire climate agenda and even its foundational values could come under threat.
Of course, the answer is not digital autarky. Instead of cutting itself off from the rest of the world, Europe must promote freedom of choice. As the European Parliament made clear in its January 22 resolution on European technological sovereignty and digital infrastructure, the first step is to map the EU’s dependencies. Following that operational challenge for the bloc, the next is to develop alternative architectures: modular, interoperable, locally governed data ecosystems that can ensure continuity for scientific and industrial uses.
Data sovereignty is more than a political slogan; it could determine Europe’s ability to produce its own knowledge. To remain a leading scientific and industrial force in climate policy, renewable energy, and sustainability, the EU must secure access to the underlying data and analytics. That requires taking Ratso’s warning seriously.
