AI governance is the new climate governance – what does this mean for Europe?

Recently, the UN’s Independent International Scientific Panel on Artificial Intelligence warned that “the traditional model of safeguarding is unravelling”. It described AI agents bypassing test controls, coordinating across systems and concealing their actions. US AI tech tycoons are calling for more regulation, AI researchers are quitting in droves, citing concerns about the consequences of AI developments. The shift from static models to agentic systems – capable of acting autonomously, at speed and at scale – is raising new questions about how these technologies can be governed.

Yet outside the inner circle of experts, the reaction remains eerily quiet. Be it shock or helplessness, the muted reaction echoes the early days of climate policy. The parallels between these two transformative challenges hold lessons Europe cannot afford to ignore.

Four striking parallels

Complexity and uncertainty. Climate governance touches atmospheric science, economics, energy and law; agentic AI spans machine learning, cybersecurity and liability. Both involve competing interests, and their politics are often driven by competition and power rather than equity or fairness. Their risks also feel abstract when weighed against everyday convenience and established behavioural habits. Both require anticipation of the future to drive action today, yet in both, jargon and uncertainty let specialists dominate the discourse while public buy-in and democratic oversight lag behind.

Global collective struggle. Emissions occur anywhere and affect everyone; AI agents cross borders instantly and compute power is in global demand. Both are global issues requiring global answers. Climate policy has accounting and target setting standards defining a common framework, but voluntary national pledges and weak enforcement mechanisms render it largely ineffective. AI governance is nascent, without any global governance framework and already fragmenting between Europe, the United States and Asia-Pacific, causing regulatory arbitrage. On both topics, geopolitical rivalry and weakened multilateral institutions leave each actor convinced it cannot make a difference alone.

Systemic, high-stakes impacts. Unmitigated global warming threatens the provision of fundamental human needs, like food and water security and can lead to social unrest and large-scale migration. Agentic and interacting AI agents can unlock a tremendous increase in productivity increase, both for life-improving and for life-threatening purposes. Regardless of how we assess the dangers of AI, even benign applications will touch millions of lives as they reshape labour markets, particularly in entry- and middle-level roles. Already today, EU hiring rates are slowing and underemployment indicators seem to be on the rise, indicating more precarious and low-level jobs for graduates. In short, both climate and AI carry tail risks with deep distributional consequences in societies, with consequences for political systems and populations.

Winners and losers. Both transitions create winners and losers, and the picture of who will be on either side is slowly emerging. Smallholder farmers, for example, have long been the prime example for potential climate losers. However, increasingly widespread climate events – think wildfires in California, France, and Spain – show that everyone can be potentially vulnerable. Especially Europe’s urban populations and social housing tenants are less able to adapt or to afford new equipment for heating and cooling. Yet, policy is slow to turn the lessons into action and provide targeted support or reality-checked solutions.

Winners are easier to identify as markets are moving: Clean transition tech and early AI movers are capturing huge opportunities, while some of Europe’s leading industrial sectors fall behind. Meanwhile, the sense of unease about unchecked and concentrated corporate and geopolitical power is growing and very little is discussed about what happens if or when the huge AI finance bubble bursts. Even less has been done – in either the AI or the climate debate – to bring the broader population along, shape a vision of the future and credibly explain the changes ahead. Thus, the question of who gains and who pays, and what is needed to balance impacts and shore up equity in societies, escapes democratic debate.

…and where the differences lie.

Whilst there are similarities between climate and AI governance conundrums, there are also significant differences: AI developments can unfold in weeks rather than decades, and damages such as mass disinformation or autonomous cyberattacks can affect a society more broadly than the geographically diffuse climate impacts experience today. On the other side, its capabilities and implications are far harder to measure than tonnes of CO₂, and the sector lacks anything like the common methodologies and standards we know in the climate space.

But AI also offers what climate never did: concentration. Frontier capabilities sit with a handful of labs and cloud providers, and compute, data, model weights and capital flows are relatively well-defined chokepoints. Currently, this contributes to a sense of dealing with a set of proprietary ‘black boxes’ but they are sufficiently concentrated for regulators to reach directly and in a targeted way, if they wanted.

Europe needs leadership, not just rulebooks

It’s clear that Europe can’t afford to sit on its hands when it comes to shaping the future of AI and climate governance. The stakes are too high. Although Europe likes to make itself small, it has influence through its collective size and policy architecture – the AI Act, the Data Act, the Digital Markets and Services Acts, the Data Union Strategy and the Green Deal – to deal with both. Both agendas are under pressure in the name of simplification and a trend to market nostalgia, but we need to get past this.

It is critical that Europe acts now as AI and climate are its twin strategic challenges: they can reinforce each other if managed through proper allocation of capital, energy, land, skills and political attention, and will be won or lost together. Steering them demands strong political and regulatory leadership, and greater societal consensus in Europe. Together, these transitions could provide the impetus Europe needs to jolt itself out of national backsliding and move to deeper integration.

Here are three proposals:

1) Make the AI transition a truly European project. AI is systemic disruption that needs a systemic response. Herein lies the opportunity to shape a ‘European project’ from the start instead of retrofitting an EU lens to national approaches. A high-level ‘European AI Commission’ with the participation of business (tech and otherwise), civil society and academia should be called to provide EU-wide and strategic recommendations for policymakers on the main AI impact pathways, including work & education as well as social & financial systems.

It should have a relatively short timeline to inform the next European elections in 2028. Its task should be to provide a plan on how to shift part of the tax burden from labour towards capital, AI-compute and automation to shore up public finance and social systems whilst financing the transition. This must include thoughts on the education system that anticipates and adapts to the integration of AI, and social systems that put greater attention and targeted fiscal care to protect vulnerable groups.

2) Hold the net-zero line and bring AI in on the transition. Keep net-zero targets as the anchor, with predictable carbon pricing and de-risking instruments. New data centre placements need a strategic European lens and should commit AI tech players to bring additional renewable capacity, co-invest in grids and storage, and be sited with regard to energy, water, land and regional cohesion, in return for faster permitting. At the same time, with the additional power, land and water needs, AI infrastructure must be roped in from the start to measure and improve resource efficiency as well as invest in climate adaptation, not only for its own needs but for the landscape it operates in.

3) Build sovereignty through one industrial strategy. Industrial, energy, data and AI policy must form one strategy. Trusted data spaces, federated infrastructure for sensitive workloads, and investment in chips, cloud and AI factories should be judged by whether they keep hospitals, grids and public services secure and strengthen Europe’s clean-industrial base. Whilst Europe can’t emulate other regions’ availability of raw materials availability, it can be smart about keeping its materials in the economy and occupying strategic technologies in the energy, machinery and tech supply chain that provides it with leverage.

A narrowing window

Safeguards are not keeping pace with capabilities, and a local failure can spread across borders within hours. Europe can repeat the delays and fragmentation that have slowed climate action or show that complex global challenges can be met with coherent policy, credible incentives and political courage.

There is good reason for confidence. Europe wrote the world’s first comprehensive climate law and its first binding AI law, and the transition it championed has made solar and wind the cheapest power in much of the world. This time the starting position is even better: the leverage points in AI are few and visible, and the lessons of climate policy are already learned. By getting both transitions right, Europe can strengthen its competitiveness, resilience and capacity to shape the global rules of the future. The window is narrowing, but it is still open.

The views expressed in this #CriticalThinking article reflect those of the author(s) and not of Friends of Europe.