The Convention on Certain Conventional Weapons remains the most realistic, if imperfect, forum for negotiating limits on lethal autonomous weapons systems. In 2024 the Group of Governmental Experts (GGE) met in two formal sessions, first in March and then again in Geneva from 26 to 30 August, under a mandate to “further consider and formulate, by consensus, a set of elements of an instrument, without prejudging its nature.” These dates and the expanded three‑year mandate frame the patently urgent task before the Group: to reconcile technical reality, legal obligation, and ethical imperative within an institution designed to act by unanimity.

Progress in Geneva has been procedural and textual rather than decisional. The Chair circulated a “rolling text” that attempted to stitch together diverse proposals into candidate elements of an instrument. That document became the focus of arguments about characterisation, scope, and, crucially, the word “lethal.” Over the course of the March session and into the August session the rolling text was revised repeatedly; by late August versions explicitly engaged with competing understandings of “lethal” and with divergent state preferences on whether the two‑tier model of prohibition plus regulation is the right way forward. The appearance of updated rolling text files on 27 August shows how much of the Group’s energy has been expended on formulation rather than negotiation of definitive commitments.

Two substantive fault lines have dominated the talks. First is the semantic and legal debate about characterisation: should the instrument target “lethal autonomous weapons systems” specifically, or a broader class of autonomous weapons irrespective of whether their effects are lethal? Some governments insist that “lethal” is intrinsic to the GGE mandate and must be retained; others argue it is legally unnecessary and potentially misleading because existing international humanitarian law applies regardless of a weapon’s label. Second is the question of where human judgment must remain. Concepts such as “meaningful human control” and “context‑appropriate human judgment” have been advanced, but they are slippery. The rolling text has tried to capture this ambiguity; states remain divided about operational thresholds that would make an automated function unacceptable versus those that would be regulated. These divisions are not merely lexical. They map onto divergent military doctrines, industrial interests, and geopolitical strategies.

The wider international context compounds the technical debate. In mid‑2024 the UN Secretary‑General compiled a report on autonomous weapons systems that synthesised more than eighty submissions from states, civil society, industry, and experts. That report has been used by many delegations to press the case that there is growing international support for a legally binding instrument, and it has sharpened messaging from proponents of prohibition or strict regulation. Civil society and research organisations likewise documented broad backing for some form of treaty action in the submissions solicited by the Secretary‑General. These inputs increased pressure on the GGE but they did not erase the structural limits of a consensus body confronting technologically and politically charged questions.

The pragmatic consequence is familiar: incrementalism. The GGE is refining definitions, collecting best practices on weapons review, and considering compliance measures, but it is not yet negotiating an operative, enforceable protocol. States favourable to rapid, legally binding measures pressed for clearer prohibitions and temporal or geographic limits on autonomous use. States more sceptical of new legal constraints prioritized safeguards, transparency measures, and national implementation approaches that preserve operational freedom. The result is a slow churn of text and a widening gulf between those who believe the CCW can produce a binding instrument in this mandate period and those content to treat the GGE as a venue for continuing dialogue.

From the perspective of strategy and robotics, this stalemate is not merely bureaucratic pedantry. Autonomy and AI capabilities are diffusing quickly. When governance lags, power asymmetries grow and norms ossify around contested practices. The moral calculus is acute: delegating life‑and‑death judgments to algorithmic systems without robust legal guardrails degrades both accountability and human dignity. The GGE must choose whether it will convert its rolling text into a negotiating mandate or allow the CCW’s requirement for consensus to dilute the project into a set of nonbinding recommendations. The former demands political courage and risk; the latter buys immediate comfort at the expense of future harms.

What should observers and practitioners watch for in the coming months? First, whether the Chair can synthesize the rolling text into a narrowed set of elements that a clear majority can endorse as the basis for negotiations. Second, whether states that have hitherto resisted prohibition proposals will accept tighter language on unacceptable systems, or continue to insist on national regulatory approaches. Third, whether momentum from the General Assembly submissions and the Secretary‑General’s report translates into institutional pressure—via the General Assembly or coalitions of the willing—to move beyond the CCW consensus constraint. Each of these pathways carries risks and trade‑offs, but the alternative is drift.

Philosophically, the LAWS debate is a test of whether international law can remain prophylactic in an age of accelerating automation. If the international community waits until the technology is ubiquitous, the choice will no longer be about abstract principle but about reversing hardened military practices. The GGE’s rolling text is a necessary step. It will not be sufficient. To convert scholarly concern and civil society urgency into legal constraint, states will need to accept normative choices that bind them operationally. That is uncomfortable. It is also necessary.

Until then, the work in Geneva will continue to reflect the best and worst of multilateralism: a conscientious effort to understand complexity, weighed down by institutional incentives to avoid binding commitments. The only honest prognosis is conditional: progress is possible, but only if a critical mass of states treats the rolling text as a starting point for negotiation rather than as a refuge from decision.