The law of armed conflict was drafted for a world of soldiers, front lines and decisions taken in the heat of battle. It now confronts machines that loiter for hours over a village, weapons that may one day choose their own targets, and software that ranks human beings for life or death. The question is no longer hypothetical: can rules written in the aftermath of the Second World War still govern the wars of the twenty-first century?
International humanitarian law (IHL) — sometimes called the law of war — does not ask whether a conflict is just. It asks how those who fight must behave. Its central bargain is a sober one: war may be lawful, but it is never lawless. The difficulty today is that the technologies reshaping the battlefield were unimaginable when the governing texts were agreed, and the gap between the letter of the law and the reality of remote, automated violence is widening.
This briefing sets out the foundations of that law, examines how armed drones already strain it, turns to the harder frontier of autonomous and AI-assisted weapons, and assesses whether the international community is capable of keeping pace. It is analysis, not advocacy: our aim is to map the legal terrain, not to take sides in the political disputes that surround it.
The foundations of the law of war
Modern IHL rests primarily on the four Geneva Conventions of 1949 and their Additional Protocols of 1977, together with a body of customary law that binds states regardless of which treaties they have ratified. The Conventions emerged from the devastation of total war and were designed above all to protect those who are not, or are no longer, fighting: civilians, the wounded, prisoners of war.
Four principles do most of the work. Distinction requires parties to distinguish at all times between combatants and civilians, and to direct attacks only at the former. Proportionality prohibits attacks expected to cause civilian harm excessive in relation to the concrete military advantage anticipated. Precautions oblige attackers to take all feasible steps to minimise harm — verifying targets, choosing means and timing with care, giving warnings where possible. And the principle of humanity forbids weapons and methods that cause superfluous injury or unnecessary suffering. These principles are deliberately framed in human terms: they assume a person capable of judgement is making the decision to strike.
How armed drones strain the rules
Remotely piloted aircraft have been a fixture of armed conflict for roughly two decades, and they do not, in themselves, break IHL. A drone is a delivery system; the principles of distinction and proportionality apply to a strike whether the pilot sits in the cockpit or in a control room thousands of miles away. In some respects the technology can aid compliance, offering long surveillance, high-resolution imagery and the option to abort.
Yet the practice of drone warfare has raised persistent and unresolved questions. The phenomenon of remote killing — lethal force applied by operators who face no physical risk, often far from any recognised battlefield — has prompted debate about where the law of armed conflict ends and the stricter rules governing law enforcement begin. So-called "signature strikes", reportedly directed at patterns of behaviour rather than identified individuals, sit uneasily with the requirement to verify that a target is in fact a lawful objective. And when civilian casualties result, the question of who is accountable — the operator, the commander, the analyst who supplied the intelligence, the state — is rarely answered cleanly. These are difficulties of application rather than of principle, but they have proved stubborn, and reliable independent figures on civilian harm remain contested and incomplete.
The machine does not extend the trigger so much as the chain of hands upon it — and a chain of hands can become a chain in which no single hand is responsible.
The harder frontier of autonomous weapons
If drones complicate the existing framework, lethal autonomous weapons systems (LAWS) and AI-assisted targeting threaten to unsettle its foundations. A genuinely autonomous weapon would, once activated, select and engage targets without further human intervention. AI decision-support tools, already reportedly in use in some conflicts, stop short of that — but they sift vast quantities of data to recommend targets at a speed no human team could match, and the line between recommending and deciding can blur under operational pressure.
The central debate concerns "meaningful human control": the proposition, widely supported though not yet settled in law, that decisions to use lethal force must remain subject to genuine human judgement rather than delegated to an algorithm. Closely related is the accountability gap. IHL assumes a responsible human actor; if an algorithm selects a target and the result is an unlawful death, it is far from clear who bears criminal responsibility. A machine cannot be tried, and stretching command responsibility to cover the unpredictable behaviour of a learning system is legally untested. Compounding both concerns are speed and error: systems that act in fractions of a second leave little room to call off a mistake, and machine perception can fail in ways that are opaque, hard to audit and difficult to anticipate.
When an algorithm rather than a person selects who is struck, IHL's assumptions begin to fray. The recurring legal questions include:
- Can a weapon system make the qualitative judgement that distinction and proportionality require?
- Where lethal force is delegated to software, who exercises the "meaningful human control" the law is increasingly said to demand?
- If an autonomous system kills unlawfully, who is criminally responsible — operator, commander, programmer, manufacturer, or no one?
- How can a strike decision taken in milliseconds by an opaque model be reviewed, explained or audited after the fact?
- Does a state's duty to conduct a legal review of new weapons adequately capture systems that learn and change in the field?
The state of international regulation
For more than a decade, the principal forum for these questions has been the United Nations Convention on Certain Conventional Weapons (CCW) in Geneva, where a group of governmental experts has discussed autonomous weapons since the mid-2010s. Those discussions have produced sets of guiding principles — reaffirming, among other things, that IHL applies fully to such systems and that human responsibility must be retained — but they operate by consensus, which has made binding commitments elusive.
Outside that forum, a growing coalition of states, international organisations and civil-society groups has called for a dedicated treaty: one that would prohibit weapons operating wholly outside human control and regulate the remainder. The United Nations Secretary-General and the International Committee of the Red Cross have urged states to agree new legally binding rules, and the General Assembly has begun to take up the issue. Yet several militarily significant states remain cautious about binding constraints, preferring non-binding standards or arguing that existing law suffices. The result, as of mid-2026, is movement without resolution: widespread agreement that something must be done, and persistent disagreement about what.
Can old law govern new wars?
It would be a mistake to conclude that the Geneva Conventions are obsolete. Their principles were drafted in general terms precisely so that they could outlast the weapons of any single era, and the prevailing view among states and scholars is that IHL applies to drones and autonomous systems just as it does to any other means of warfare. The harder truth is that applying old principles to new technology is not automatic: distinction and proportionality presuppose human judgement, and a rule is only as strong as the mechanisms that enforce it.
The realistic path is neither wholesale replacement nor complacent reliance on texts alone. It lies in clarifying how existing principles bind emerging systems, insisting on meaningful human control, requiring rigorous legal review of new weapons, and building accountability mechanisms equal to the speed and opacity of the technology. Whether the international community can do so before the technology outruns the law is, for now, an open question — and one that deserves clear-eyed analysis rather than either alarm or dismissal.
- International humanitarian law rests on distinction, proportionality, precautions and humanity — principles that assume a human being is making the decision to strike.
- Armed drones do not break these rules in themselves, but remote killing, "signature strikes" and diffuse accountability have strained their application.
- Autonomous weapons and AI-assisted targeting raise harder questions: meaningful human control, an accountability gap when an algorithm selects a target, and the risks of speed and error.
- Regulation is advancing slowly through the Convention on Certain Conventional Weapons and growing calls for a treaty, but binding rules remain elusive as of mid-2026.
- The Geneva framework is not obsolete, but applying century-old principles to new technology demands deliberate clarification, control and enforcement.
How Crejj & Partners can help
Our International & Public Law commentary tracks the legal dimensions of a fast-changing security landscape, including the regulation of emerging military technology. We advise clients — from technology developers and defence-adjacent businesses to investors and institutions — on emerging-technology legal risk, export and sanctions exposure, compliance frameworks, and the reputational and contractual questions that arise when products touch the law of armed conflict. Where matters require litigation or specialist representation, we work alongside the appropriate experts. If you need to understand how this evolving body of law may bear on your organisation, we are glad to talk.
This article is provided for general information and analysis only. It is not legal advice, does not constitute political endorsement of any party, and does not create a solicitor–client relationship. Descriptions reflect publicly reported information as of mid-2026 and may since have changed. Crejj & Partners is a fictional firm presented for illustrative purposes on this website.