Code as Law? Why Arms Control is Failing the Governance of Autonomous Weapons

As autonomous weapons rapidly gain momentum, treaty negotiations in diplomatic forums like the Convention on Certain Conventional Weapons (CCW) are at risk of being replaced by norms emerging from secret state practices and determined by corporate priorities. The decision on whether to start negotiations on a treaty governing autonomous weapons systems (AWS) at the 2026 CCW Review Conference will serve as a test for where governing norms for critical technologies come from in the future.

Militaries all over the world are locked in a high-stakes race to integrate Artificial Intelligence (AI) into weapon capabilities and decision-making. Algorithmic targeting support systems like Lavender generate targets at speeds no human can match. Loitering munitions and drones with autonomous terminal guidance reportedly «finish the job», even when cut off from their operator in the electronic warfare environment. Cheap autonomous weapons systems (AWS) routinely destroy multi-million dollar military equipment and are quickly replaced – resulting in a new, asymmetric economic frontline where shooting down a drone frequently costs more than manufacturing one. Autonomy is no longer a thing of the future. It is rewriting military doctrine in real time on battlefields, from Gaza to Ukraine. 

Strategic Reality vs. Abstract Debates

Very much in contrast to the reality of AWS on today’s battlefields, debates on binding rules for their development and use remain tragically abstract. It took the Convention on Certain Conventional Weapons (CCW) Group of Governmental Experts on Lethal Autonomous Weapons Systems (GGE) 12 years alone just to agree on how to define AWS and the resulting framework does not sufficiently address concerns surrounding a system’s predictability, reliability, and explainability. While the GGE reaffirmed human control as prerequisite for compliance with International Humnitarian Law (IHL), powerful states developing AWS have steadily watered down its meaning. Notably, human authorization as a requirement for alterations of a system’s target parameters or mission did not make it into the GGE’s final recommendation. 
Most importantly, the GGE stopped short of recommending a prohibition norm – effectively greenlighting the development and use of AWS. The history of arms control bears a predictable lesson: innovation almost always outpaces collective efforts for governance. With great power conflict on the horizon, neither the US nor China are willing to negotiate away battlefield advantages. For regional powers like Iran, the inverted economics of autonomous warfare offer an irresistible strategic opportunity to offset power asymmetries. And a general reluctance among states to rest their security on rules that do not exist yet or are not considered binding by adversaries ultimately ensures that bans and restrictions are mostly secured for weapons which are obsolete, strategically not advantageous, or not yet real – none of which apply to AWS.

Drivers of Paralysis: Opaque State Practices and Corporate Capture

Yet efforts to regulate AWS and military AI are further crippled by a profound shift in how international norms develop. From remote drone strikes to targeting decisions based on generative learning algorithms – states’ secret operational practices since 9/11 have reshaped fundamental norms like human oversight. Though not necessarily compliant with IHL, these novel norms satisfy standards of „procedural-organizational appropriateness“, such as efficiency, and manage to appear normal and legitimate when spreading over time. Shielded behind the opaque curtain of states’ security agencies, autonomous technologies enable practices which gradually change the essence of existing IHL norms traditionally guiding, for example, decisions on distinction and proportionality. 

Deliberations on new binding regulations on the other hand are obstructed by the growing influence of corporations in the rulemaking on AWS and military AI. Governments routinely grant tech leaders memberships in closed-door advisory committees like the EU’s Global Tech Panel where they set agendas and steer discourse. Even more directly, corporations shape the boundaries of regulation through the engineering choices embedded in their hard- and software. By declining to recommend a two-tier prohibition covering not only the use but the development of AWS and military AI, the GGE kicked the regulatory can down the road.  For tech companies capitalizing on lucrative defense contracts in an unregulated frontier, the GGE’s deadlock is not a diplomatic failure, it is a business opportunity. 

Bridging Innovation and Trust

Whether states decide to launch negotiations for a binding treaty at the CCW Review Conference starting this week will send a signal reverberating far beyond Geneva. Allowing for governing norms to emerge from secret practices and for technical attributes to privilege policy options that serve corporate interests, means that epistemic authority on AWS and military AI will be transferred from the public to boardrooms. The faultlines are clear: an overwhelming majority of states supporting negotiations faces a small group of states including the US, Russia, and Israel that put up “smokescreens to keep [AWS] unregulated and states unaccountable for their use.” The CCW decision will show whether the international community is ready to refuse rubber-stamping a technological fait accompli and overcome multipolarity. 

Switzerland has already positioned itself as a promoter of collaboration across the demarcation between AWS developers and the supporters of their regulation – bringing the conversation on artificial intelligence right into the heart of multilateralism at the 2027 Geneva AI Summit. And yet it has to keep evolving as mediator and host not just in humanitarian but tech diplomacy to prevent a global order defined by algorithmic momentum. With Geneva as the world’s humanitarian center of gravity and leading tech experts, Switzerland is uniquely positioned to  identify and scope scientific advances most likely to reshape warfare and embed projections on the related risks into the diplomatic conversation before disruption harms civilians.

Zum gleichen Thema