Unchecked AI growth could deepen climate chaos

Climate stability and AI safety are global public goods. No single country can secure them alone, and failures by one major actor can undermine collective efforts. The study shows that governance models shape how societies manage these shared risks, determining whether cooperation or competition dominates.

A new peer-reviewed study warns that without deliberate global governance choices, artificial intelligence (AI) could just as easily accelerate climate collapse as help prevent it.

Published in AI & Society, the study titled “Four visions on the future of global governance, AI, and climate change” examines how political institutions, technological development, and environmental trajectories may interact by 2045. The research does not attempt to forecast the future. Instead, it develops four plausible global scenarios to expose the consequences of today’s policy decisions and governance failures.

Governance decides whether AI deepens or limits climate harm

Climate stability and AI safety are global public goods. No single country can secure them alone, and failures by one major actor can undermine collective efforts. The study shows that governance models shape how societies manage these shared risks, determining whether cooperation or competition dominates.

In one scenario, the world drifts toward an anarchical, market-driven order dominated by powerful corporations and weak states. In this future, AI development accelerates without meaningful oversight, driven by competitive pressure and profit motives. Energy use from massive computing infrastructures rises sharply, intensifying climate damage. Climate responses exist, but they are uneven, privatized, and accessible only to those who can afford them. Technological solutions become commodities rather than collective safeguards, widening inequality and instability.

The authors contrast this with futures shaped by stronger coordination. In governance systems built on voluntary norms and cooperation, states restrain AI development by agreement, prioritizing shared goals such as emissions reduction, environmental justice, and social stability. While innovation slows, risks decline, and climate outcomes improve compared to unregulated acceleration. However, this approach remains fragile, vulnerable to free riders and geopolitical mistrust.

More formalized international regimes offer another pathway. In this model, binding rules, monitoring mechanisms, and enforcement structures govern both AI and climate policy. Technological progress continues at a rapid pace, but under constraints designed to prevent catastrophic misuse and environmental overreach. AI becomes a strategic tool for climate modeling, mitigation, and adaptation, while institutional oversight limits the most dangerous applications.

The most ambitious scenario imagines a world government capable of enforcing global rules. Here, coordinated authority allows AI to be deployed intensively for climate stabilization while sharply restricting harmful uses. Emissions decline toward best-case trajectories, but the cost is reduced national sovereignty and increased central control. The study does not endorse this outcome, but highlights the trade-offs inherent in any governance choice.

Across all scenarios, the message is consistent: AI does not inherently lead to sustainability or disaster. Governance determines the outcome.

AI as both climate burden and climate instrument

On one hand, large-scale AI systems consume vast amounts of electricity, water, and land. Data centers strain power grids, often relying on fossil fuel-heavy energy mixes. The rapid expansion of generative AI intensifies these pressures, raising concerns that AI growth could undermine climate goals.

On the other hand, AI offers powerful capabilities that could transform climate action. Advanced models can improve climate forecasting, optimize energy systems, support emissions tracking, and accelerate scientific discovery. AI can enhance understanding of Earth systems and help design more efficient infrastructure across sectors from agriculture to transportation.

The authors caution against technological optimism that assumes AI will automatically solve climate change. Efficiency gains often trigger rebound effects, where cheaper or faster systems lead to higher overall consumption. Without governance, AI-driven optimization can simply enable more intensive exploitation of natural systems.

In several scenarios, AI contributes to climate adaptation rather than prevention. As emissions rise, societies increasingly rely on technological fixes such as climate engineering, artificial environments, and large-scale interventions. These responses may reduce immediate harm for some populations while creating new risks, ethical dilemmas, and geopolitical tensions.

The study also highlights how AI’s climate impact extends beyond emissions. Resource extraction for hardware, water consumption for cooling, and land use for infrastructure all add to environmental stress. These material realities undermine narratives that frame AI as a purely digital or immaterial solution.

Ultimately, the research positions AI as a force multiplier. It amplifies existing political and economic dynamics. In cooperative systems, it can support collective action. In competitive or fragmented systems, it intensifies conflict, inequality, and environmental degradation.

Why today’s choices shape the world of 2045

The authors use scenario analysis to challenge current assumptions. They argue that policymakers often underestimate how quickly technological trajectories can lock in long-term consequences. Governance structures evolve slowly, while AI capabilities scale rapidly, creating a widening gap between innovation and oversight.

Governance failures already visible today could harden into systemic risks. Weak international coordination, voluntary climate commitments without enforcement, and fragmented AI regulation all point toward futures where collective action becomes harder over time. As AI becomes central to economic growth and military power, incentives to cooperate weaken further.

The research rejects fatalism. It shows that alternative pathways remain plausible if political will aligns around shared priorities. Investments in global institutions, transparency, and enforceable norms can redirect AI development toward public benefit. Climate outcomes improve not because technology advances faster, but because it is used more deliberately.

Justice and equity emerge as critical themes. In poorly governed futures, climate protection becomes a privilege rather than a right, and AI-driven solutions deepen divides between regions and populations. In more coordinated systems, burdens and benefits are distributed more evenly, though often at the cost of autonomy and local control.

The authors stress that no scenario is without trade-offs. Strong governance can constrain innovation and individual freedoms, while weak governance enables exploitation and instability. The purpose of the scenarios is not to prescribe a single solution, but to clarify what is at stake.

By 2045, the interaction between AI and climate change will reflect decisions made in the 2020s and 2030s. Choices about regulation, cooperation, and institutional design will determine whether AI becomes a destabilizing accelerant or a managed instrument for sustainability.