Asia-Pacific Journal Series – The Martens Clause: International Law’s Favourite Rule That Nobody Uses

Editors’ note: This post appears as part of a series highlighting articles in a recent volume of the Asia-Pacific Journal of International Law, on which presentations were made at a workshop hosted by National University of Singapore’s Centre for International Law. Introductory post by Samuel White, Danielle Yeow, and Jonathan Kwik can be found here.

Spend enough time at conferences on military artificial intelligence (AI) and it is almost inevitable that someone will invoke the Martens Clause.

Such conference discussions range across autonomous weapons, AI decision-support systems, accountability, cyber operations, and data poisoning. Yet regardless of where the conversation began, it seems eventually to arrive at the same destination. Whenever existing law appears uncertain, or technological developments has outpaced treaty law, the Martens Clause emerges as the answer waiting quietly in the background.

This constant return to the Martens Clause triggered me to change course in my research away from the historical triggers that created legal personhood, and to take a deep dive into the Martens Clause. This post builds upon the article that resulted from that research.

A Divide Between Theory and Practice

The Martens Clause is invoked constantly in academic discussions of emerging military technologies, yet remarkably rarely elsewhere. It occupies an increasingly prominent place in journal articles, conference papers, and expert workshops, but enjoys comparatively little attention in military manuals, operational legal advice, or the public positions adopted by States (I cannot find any reference in any State position).

More than a century after it first appeared, there remains surprisingly little State practice explaining what the Clause actually requires. It has become one of the most frequently cited concepts in academic discussions of military AI while remaining one of the least developed (or relevant) for operational lawyers.

That observation is not intended as a criticism of either community, as I sit in both. Rather, it reflects that academics and practitioners are trying to answer different questions. Military lawyers advising commanders are concerned with identifying the law as it presently exists. Academic lawyers are often interested in where the law may need to develop when confronted with technologies that existing treaties never contemplated. AI has brought those two perspectives into unusually close contact because it increasingly presents problems that fit comfortably into neither category.

A Case Study: Data Poisoning

A useful example of this phenomenon is data poisoning. Data poisoning served as the operational case study for my previously mentioned article (see also my Articles of War post from earlier this year). Instead of attacking a military system directly, an adversary quietly corrupts the data used to train another State’s AI. The operation may take place years before any armed conflict exists. No infrastructure is destroyed. No obvious damage is visible. The affected system continues to function until, perhaps during a future conflict, it begins producing unreliable intelligence assessments, flawed targeting recommendations, or compromised logistical planning. The hostile act belongs to peacetime, but its consequences emerge only in war.

This is precisely the kind of problem that encourages lawyers to reach for the Martens Clause. International humanitarian law is highly developed once armed conflict exists, while international human rights law governs much of what occurs outside it. Grey-zone cyber operations such as data poisoning deliberately blur that distinction. They exploit the seams between legal regimes rather than operating comfortably within either of them. It is hardly surprising, therefore, that scholars repeatedly return to a provision that the International Court of Justice has described as “an effective means of addressing the rapid evolution of military technology.”

Can the Martens Clause Bear the Weight?

This is where the paradox emerges. The Martens Clause is often treated as though it fills the gaps left by positive law, but the Clause itself remains largely unexplored by those responsible for creating that law. There is little consistent State practice, limited operational guidance, and remarkably little agreement about what the principles of humanity or the dictates of public conscience require in concrete legal terms. In many respects, the Clause has become international humanitarian law’s preferred theoretical solution to emerging technology while remaining an underdeveloped source of practical legal obligation.

Concluding Thoughts

That gap matters. If artificial intelligence is exposing areas where treaty law remains incomplete, then simply invoking the Martens Clause is no longer enough. The more difficult question is whether the Clause is capable of performing the role increasingly assigned to it by contemporary scholarship, and if so, what that means for States, military lawyers and commanders confronting technologies that the drafters of the Geneva Conventions could scarcely have imagined.

***

Dr Samuel White is the Scientia Senior Researcher in Military Law and War Studies at UNSW Canberra (based within the Australian Defence Force Academy).

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense. 

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

Photo credit: Unsplash