Forecasting Power: AI in the Arctic

Signals from the High North

In the Arctic, the stakes of weather forecasts are uniquely high. From maritime safety to infrastructure protection to national defense, prediction is foundational to decision-making in an environment marked by rapid change and extreme conditions. Yet despite their critical importance, weather forecasts rely on sparse and dirty data, often drawn from sensors and systems operating across borders, jurisdictions, and legal regimes.

Artificial intelligence (AI) and machine learning (ML) have entered the Arctic weather enterprise with remarkable speed. These algorithms (sets of coded instructions that enable computers to learn from data) now underpin everything from sea-ice prediction and aviation guidance to logistics planning and emergency response.

They promise improved accuracy and timeliness in a region where conventional observations are limited. But their integration is not merely technical – it reshapes who controls information, who bears responsibility, and what rules apply. While the High North signals, the international community must listen.

Regulating Uncertainty

The Arctic forecasting system is embedded in a web of international arrangements. These include most fundamentally the Convention of the World Meteorological Organization (WMO) requiring open and interoperable data exchange from land, sea and space based sensors underpinned by defined authorities for forecasts and warnings.

The complex of data streams is governed by standards set by agencies such as the International Telecommunication Union (ITU) and the International Organization for Standardization (ISO), characterized by self-interested compliance. Generally applicable law, such as the UN Convention on the Law of the Sea (UNCLOS) and the UN Convention on International Civil Aviation (Chicago Convention), sits alongside specialized Arctic instruments such as the International Maritime Organization (IMO) Polar Code.

There is no unified legal regime for the Arctic weather value cycle. Rather, there is a mosaic: generally applicable international law, sectoral and geographic instruments, and evolving soft law. These interact with domestic regulations, Indigenous data rights, and commercial contracts, producing an accord built on trust more than paper.

Private Infrastructure and Public Risk

AI-enabled forecasting worldwide increasingly depends on private technology providers, and the Arctic is no exception. These firms supply the compute infrastructure and model development platforms that shape how forecasts are made and delivered. But they are accountable primarily to shareholders, subject variously to product-liability, consumer protection and professional negligence doctrines that vary across countries.

There is greater freedom to restrict data, neural weights (settings for AI/ML algorithms), or the algorithms themselves as commercial-in-confidence.

This dynamic creates a structural asymmetry. Public meteorological agencies are statutorily required to provide products and services, while private actors can withdraw services or alter access with little oversight. In the Arctic context, where safety, sovereignty, and science intersect, this imbalance introduces legal and operational vulnerabilities.

Hybrid Forecasting and Global Security

Weather in the Arctic is not neutral. The forecast is not merely an auxiliary product; it is a critical enabler of defense posture, force mobility, and situational awareness. When forecasts fail, risks to assets and personnel escalate, and the credibility of deterrence itself may erode.

If a forecast is generated by a commercial provider, complex questions of responsibility emerge. The picture is further complicated by the additional threat surfaces introduced by AI/ML: from the obvious – like data stream jamming – to the more insidious, such as model poisoning. States may find themselves dependent on tools they cannot fully interrogate, audit or regulate.

Our Common Interest

International law is not absent, but it is behind. Current instruments do not yet reflect the structural transformation introduced by AI/ML systems in Arctic forecasting.

A constitutive legal process is now unfolding. It draws from multiple sources: WMO resolutions, UNCLOS provisions, standard-setting bodies, and data protection regimes. These must converge into a coherent framework that reflects the realities of AI-mediated services, accommodates shared risks, and ensures accountability across jurisdictions and sectors.

In the absence of such coordination, the Arctic weather value cycle may fragment under geopolitical tension, vendor consolidation, or algorithmic opacity. Worse still, without proactive governance, the region could become a proving ground for an algocracy – a system where power is exercised through unaccountable technical systems rather than deliberative law.

Avoiding this future demands a renewed commitment to the common interest. That means hardening forecasting infrastructure against cyber threats, embedding reciprocity in data-sharing agreements, and clarifying liability in mixed public-private architectures. The quiet accord of voluntary compliance may need to evolve toward legal innovation that matches the technological pace.

AI may be transforming the mechanics of forecasting, but only international cooperation and normative clarity can safeguard the integrity of the system upon which lives – and peace depend.

Amanda H. Lynch is the Lindemann Distinguished Professor, Dept of Earth, Environmental and Planetary Sciences, Brown University, USA; Chair, Research Board and Co-Chair, Joint Advisory Group on AI, World Meteorological Organization

Charles H. Norchi is the Benjamin Thompson Professor of Law, University of Maine School of Law, USA