AI-Enabled Weapons Systems and the Environment: The Overlooked Costs of AI-Driven Warfare

Introduction

The connection between AI-enabled weapons techniques and environmental hurt grown in significance in modern society. These two areas fall inside fields which have attracted growing world consideration: on the one hand, the safety of the setting, and on the opposite, the fast growth of synthetic intelligence (AI), that means the know-how that permits machines to simulate human studying, comprehension, downside fixing, determination making, creativity and autonomy.

AI-enabled weapons techniques are usually outlined as weapons that depend on machine studying algorithms, which can embody deep studying methods, to carry out crucial features. Examples embody a weapon that makes use of AI for object recognition to tell the concentrating on course of, a system that identifies incoming missiles and doubtlessly engages them autonomously, or a decision-support system that advises a unit chief on attainable tactical choices. On the similar time, autonomous weapons techniques are typically described as weapons able to choosing and making use of power to targets with out human intervention.

Some AI-enabled weapons techniques additionally fall throughout the class of autonomous weapons techniques, notably when they’re able to independently choosing and fascinating targets, as within the case of deadly autonomous weapons techniques (LAWS). It is very important observe that whereas LAWS able to impartial concentrating on are AI-enabled, not all AI-enabled weapons techniques qualify as LAWS; many function with human oversight or restricted autonomy, performing complicated features with out totally impartial concentrating on. Some function utilizing pre-programmed, rule-based logic. For example, a few of the older techniques such because the Phalanx Shut-In Weapon System operate based mostly on reactive suggestions inside strict, predefined parameters, with out the complicated decision-making capabilities of AI.

This text, nevertheless, focuses on AI-enabled weapons techniques: these incorporating machine studying or different algorithms that require intensive knowledge processing and energy-intensive computational coaching. These techniques elevate related environmental issues that transcend these related to conventional weapon applied sciences. Certainly, AI-enabled weapons techniques can come into battle with the setting, particularly in the course of the extraction of minerals and in the course of the AI coaching section, that means the method by which AI algorithms are refined by means of simulations and knowledge evaluation previous to deployment. As soon as educated, some AI-enabled weapons techniques might doubtlessly be used – straight or not directly – to trigger critical environmental hurt, whether or not deliberately or unintentionally. States, in line with Article 36 of Protocol I to the Geneva Conventions, are underneath an “obligation to find out whether or not [AI-enabled weapons systems] employment would, in some or all circumstances, be prohibited” by guidelines of worldwide legislation.

Given the increasing reliance on such applied sciences by armed forces worldwide, it’s essential to contemplate the potential environmental implications arising from mineral extraction, AI coaching, and subsequent deployment of those techniques.

The Extractive Dimension of the AI Arms Race: A Additional Concern for the Setting

A primary crucial side of the worldwide race to develop AI-enabled weapons techniques lies within the extraction of minerals. The accelerating push by States to develop and deploy such applied sciences has pushed an growing demand for crucial minerals, usually sourced from the African continent, that function the spine and constructing blocks of AI {hardware}. The African Fee on Human and Peoples’ Rights has underscored the necessity to respect and defend elementary human rights within the extraction of assets destined for AI navy know-how growth.

But, from a global legislation perspective, there stays a big regulatory hole: present authorized frameworks provide restricted safety and enforcement mechanisms to stop the exploitation and expropriation of pure assets, notably in relation to cobalt, lithium, and uncommon earth components. In follow, the extraction and processing of those crucial minerals usually entail large-scale mining operations, resulting in deforestation, water contamination, and long-term ecological air pollution.

As famous in a latest skilled coverage observe,

the idea of accountable AI within the navy area should prolong past mere compliance with IHL requirements on the level of deployment. It should additionally embody the moral and authorized concerns on the developmental stage.

This developmental stage ought to be understood as encompassing each the extractive processes that make AI-enabled weapons techniques development attainable and in addition its coaching. On this regard, the institution in 2024 of the UN Secretary-Basic’s Panel on Crucial Power Transition Minerals represents a optimistic growth. Amongst its guiding rules, the Panel emphasised that “the integrity of the planet, its setting and biodiversity should be safeguarded”.

Nevertheless, one of many central issues surrounding AI-enabled weapons techniques lies within the highly effective pursuits of States and personal firms driving mineral extraction. Imposing stricter limitations on these actions, ideally in a global treaty, is crucial to safeguard each the environmental and human well being.

The Environmental Affect of AI-Enabled Weapons Techniques Coaching

The coaching of AI-enabled weapons techniques raises critical environmental issues resulting from their vital carbon footprint. This derives primarily from the fixed vitality consumption required to make sure their operational responsiveness, in addition to the intensive use of water (typically even potable water) for cooling functions.

Within the civilian sector, research have already demonstrated the environmental and well being impacts of information facilities, notably on susceptible or deprived communities. The same threat arises within the navy context: as an growing variety of states—and presumably in some circumstances, paramilitary teams—might achieve entry to such applied sciences, invoking deterrence rationales similar to these used for nuclear weapons, the continual coaching of those techniques seems to be a everlasting necessity.

This ongoing cycle of energy-intensive coaching might end in environmental and well being penalties similar to, if not larger than, these already noticed within the civilian sphere with the proliferation of information facilities and related air pollution dangers. This presents a paradox in a world the place the discount of carbon emissions is (or, given present resistance, ought to be) a world precedence.

An preliminary sensible step towards addressing this problem might contain limiting the environmental impression of AI-enabled weapons techniques coaching by means of each preventive and repressive measures. The primary level issues the prohibition of AI-enabled weapons techniques that may additionally fall into the LAWS class and have a sure diploma of autonomy, as advised by the so-called two-tier strategy. Such a measure would assist scale back the general variety of techniques out there and, consequently, the quantity of coaching required, whereas additionally reducing the danger of unintended or unauthorized environmental hurt.

The second level includes establishing particular limits on emissions, vitality use, and water consumption related to AI-enabled weapons techniques coaching. Such measures wouldn’t solely assist regulate using scarce assets, similar to water in sure areas, but in addition scale back air air pollution, which poses dangers to the setting and to human well being. Ideally, these provisions ought to be integrated right into a future worldwide treaty on Deadly Autonomous Weapons Techniques, to be concluded by the tip of 2026 as known as for by António Guterres, Secretary-Basic of the United Nations, and Mirjana Spoljaric, President of the Worldwide Committee of the Purple Cross. In doing so, it is very important acknowledge that these limits would particularly goal AI-enabled techniques, moderately than these LAWS that function with out AI, for which various regulatory measures would have to be developed.

Lastly, a global sanctioning mechanism ought to be established to penalize actors who exceed these outlined thresholds. The absence of efficient punitive penalties would make enforcement extraordinarily troublesome in follow. That is evident, as an example, within the non-binding nature of the UN Guiding Rules on Enterprise and Human Rights (UNGPs), which apply to enterprise actions. In such circumstances, firms are sometimes insufficiently deterred from polluting the setting, main (amongst different elements) to continued environmental air pollution. Consequently, a future treaty mustn’t solely embody emission limits for AI-enabled weapons techniques coaching but in addition present for clear accountability measures in circumstances of non-compliance, moderately than relying solely on formal thresholds. It ought to additional guarantee significant session with communities straight affected by AI-enabled weapons techniques coaching and assure that, each time air pollution or environmental hurt happens, applicable remediation processes are promptly undertaken.

The Deployment of AI-Enabled Weapons Techniques and the Environmental Affect Linked to the Coaching

A second space of concern, each within the current and particularly for the long run, pertains to using AI-enabled weapons techniques as soon as they’ve been educated and are prepared for real-world deployment. Inside this context, two distinct situations might be recognized through which environmental hurt might happen.

The primary includes conditions the place the first targets are combatants, infrastructure, or navy property, whether or not on land, at sea, or within the air. In such circumstances, the setting might undergo harm as a secondary or collateral consequence, moderately than because the meant object of assault. The second situation, nevertheless, represents what could also be regarded—notably from the standpoint of environmental legislation—as an much more critical menace: the deliberate use of AI-enabled weapons techniques to destroy the setting, or elements thereof, inside a given territory. In such situations, the motivations might fluctuate broadly and should even embody financial concerns, such because the intentional devastation of a specific space to allow its subsequent exploitation for actions yielding substantial monetary achieve.

No matter which of the 2 situations happen, an essential widespread issue to contemplate is who conducts the coaching for AI-enabled weapons techniques. If the coaching course of is outsourced to personal actors—even in collaboration with the state—these entities might conduct it according to their very own pursuits, doubtlessly biasing AI-enabled weapons techniques towards particular outcomes. Amongst such pursuits is perhaps the deliberate degradation or destruction of sure environments, with the intention of repurposing these areas for future financial or industrial actions able to producing substantial income. This situation just isn’t purely speculative, as quite a few circumstances of AI coaching in different sectors have already led to ethically questionable outcomes. In such a context, AI-enabled weapons techniques might be educated with parameters that implicitly or explicitly embody environmental destruction as soon as deployed, notably in conditions the place human oversight is restricted or solely absent.

This raises the relevance of two key ideas. The primary issues the UNGPs, which, though non-binding, ought to function a framework for accountable non-public enterprise conduct. The problem, nevertheless, lies in reconciling these non-binding requirements with highly effective financial—and, at occasions, navy—pursuits which are usually mistakenly prioritized over environmental safety. The second issues the necessity, not solely in relation to environmental use however to all types of deployment, to determine clear restrictions in any future worldwide treaty on using AI-enabled weapons techniques (particularly some varieties of LAWS) that function with minimal or no human oversight.

On this regard, the Group of Governmental Consultants (GGE) on Deadly Autonomous Weapons Techniques, as mirrored within the Rolling Textual content of 12 Might 2025 following the March 2025 session, and in subsequent discussions in the course of the September 2025 session, emphasised that States ought to guarantee context-appropriate human judgment and management over using LAWS. A revised model of the Rolling Textual content, anticipated quickly, will combine the proposals and areas of convergence recognized in September. The GGE, in March 2025, underscored that

LAWS are operated underneath a accountable chain of command and management. This contains guaranteeing evaluation of authorized obligations and moral concerns by a human, particularly, with regard to the consequences of the choice and engagement features,

a normal of conduct that may and may embody consideration of potential environmental harm.

Conclusions

What emerges from this evaluation is that, as of 2025, the extraction course of and the coaching of AI-enabled weapons techniques signify an element that should be taken significantly in relation to each environmental well-being and human well being. Ignoring these points or failing to include limits in a global treaty concerning mineral extraction, vitality consumption, water use, and ensuing emissions, dangers contributing to a world improve in air pollution, straight contradicting ongoing efforts to scale back environmental hurt.

Historical past has proven that environmental issues are sometimes addressed solely after harm has occurred, ceaselessly leading to vital penalties for each populations and ecosystems. Because of this, it’s important within the coming years to develop laws able to attaining broad consensus on threshold limits to be included in such a treaty. The danger of omitting such limits solely as a result of incapability to succeed in settlement is unacceptably excessive.

Furthermore, the treaty ought to take note of the dangers related to outsourcing the coaching of AI-enabled weapons techniques to personal actors. Establishing clear guidelines, once more with broad consensus, would assist forestall opaque or improper practices which are troublesome to detect after the actual fact and, most significantly, extraordinarily difficult to remediate as soon as environmental and human well being hurt has occurred.