A Better Path to a Treaty Banning ‘Killer Robots’ Has Just Been Cleared

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on LinkedIn (Opens in new window)
• Click to share on Facebook (Opens in new window)
• Click to share on Email (Opens in new window)

By some media accounts, the recent Convention on Conventional Weapons Review Conference was a colossal disappointment for advocates of a treaty ban on autonomous weapons systems. After 10 years of calls for a ban on so-called killer robots—including powerful arguments against their use from scientists, scholars, engineers, Nobel laureates and a wide-ranging network of civil society organizations—governments at the RevCon, as the conference is known to participants, could come up with little more than an agreement to keep talking. Fortune magazine reported that “the world just blew a major opportunity.” In reality, however, the outcome at the RevCon is neither surprising nor troubling. Contrary to some arguments, it will actually benefit the campaigners against killer robots, rather than deal them a blow. In fact, the only real surprise here is that killer robots are still being discussed at all in the context of the Convention on Conventional Weapons, or CCW, since other venues are more conducive to rapid, robust norm-building to prohibit weaponized AI. The CCW is a multilateral treaty designed to regulate weapons deemed to be inhumane. Every five years, state parties to the treaty meet in Geneva to discuss whether it should be updated, and in theory it is possible to address certain new or newly problematic weapons through adding an Additional Protocol to the treaty. But very few weapons have ever been regulated, much less any completely banned, under its auspices. The key reason is that, like many United Nations treaties, the CCW operates by consensus. This means that states who prefer to keep or develop weapons seen as inhumane by others have the ability to simply veto or stall progress on additions to the treaty. And that means that any agreements that do come out of these Review Conferences are likely to be weak and watered down, catering to the lowest common normative denominator. For example, the Additional Protocol on incendiary weapons doesn’t ban burning soldiers alive as inherently inhumane; it only rules out firebombing civilian areas. The most powerful, comprehensive weapons bans we now have today—on landmines, cluster munitions and nuclear weapons—originated not from these kind of U.N.-based consensus-driven negotiations, but through the efforts of consortiums of small and middle power states, flanked by NGO networks. As political scientist Patrick Cottrell has documented, a key step in ultimately achieving the first two of these bans was abandoning the CCW process and pushing for a stand-alone treaty instead. Hosted in major middle power capitals, these single-issue treaty processes enabled norm champions not only to shape the agenda, as political scientist Richard Price notes, but also to set up their own majoritarian procedural rules. Indeed, norm champions could and did make the very process of extending invitations to participate in the forums an exercise in moral suasion.

The outcome at the CCW RevCon bodes well for campaigners, who have always hoped to remove the killer robot ban process to a friendlier forum.

Despite ultimately leaving out a significant number of states whose compliance would be desired, these treaties have nonetheless created powerful norms that have dramatically altered even the behavior of nonsignatories. The new Nuclear Weapons Ban Treaty is similar. The nuclear powers reject it, but campaigners argue that the existence of the treaty itself and the process by which it was agreed upon have strengthened the nuclear taboo. It is perhaps no coincidence, for example, that the five permanent U.N. Security Council members recently issued a rare joint declaration that “a nuclear war cannot be won and must never be fought,” while pledging to avoid the use and spread of nuclear weapons. This means that the admittedly flaccid outcome at the CCW RevCon nevertheless bodes well for campaigners, who have always hoped to remove the killer robot ban process to a friendlier forum as a step toward swifter, less contested, more robust treaty-making. So why have NGO campaigns engaged the CCW at all, if their ideas were sure to fail in that venue? The key reason is that campaigners view the CCW as a venue for recruiting a coalition of like-minded states to a cause, rather than a forum to convince the veto players. Stand-alone treaty-making conferences get organized and hosted by states, not by global civil society NGOs. To attract sufficient state support for such a separate conference, and a lead state to champion and fund the cause, NGOs must begin by meeting states where they are. And in this domain, that is the CCW Review Conference. In the humanitarian disarmament context, the RevCon is the focusing event where the global diplomatic community of practice around inhumane weapons can best be found in one place. It is in the side events, preparatory committees and plenary meetings at these conferences that new issues are articulated and understandings carved out informally, quite aside from what states are willing to put in a formal text. Here, activists also identify and befriend governmental allies in the diplomatic corps, take stock of the prevailing counterarguments and sharpen their own collective understanding. Even more importantly, multilateral conferences are places where interpersonal connections and relationships between campaigners and delegates are developed, generating the social lubricant that greases the wheels of norm-building, persuasion and common cause. For all these reasons, the past 10 years of work at the CCW has not been at all a failure for NGOs pushing for a ban on killer robots, but rather a necessary step in a process that was always likely to have been consummated outside the CCW framework. If the goal was to give the CCW a chance to fail, then NGOs have succeeded at all of the above, paving the way to now split off to a state-led treaty-making process. And research confirms that it is in part failures of multilateral processes that galvanize state support and stigmatize veto players. In fact, as political scientist Margarita Petrova writes, the stigmatization of the veto players is a key step in successful norm-building. If there is anything surprising in these developments, it is that the campaign to ban autonomous weapons systems didn’t reach this moment sooner, after a similar result at the 2016 Review Conference. After all, the Ottawa process that ultimately led to the ban on landmines took off after just one failed effort at the 1996 CCW Review Conference. Similarly, campaigners to ban cluster munitions moved to the Norwegian-led Oslo Process after only one effort at the CCW in 2006. With killer robots, campaigners have now stuck it out for two CCW conference cycles spanning 10 years. The answer here lies in the fact that there are at least three things that make the campaign to ban killer robots especially challenging for advocates. The first had nothing to do with the politics of disarmament, but rather with the COVID-19 pandemic. The disease and resulting patchwork of virtual and hybrid approaches to global gatherings has complicated travel to, and social interaction at, major international forums, as well as the ability of civil society to gather for their separate focusing events. While diplomats and campaigners alike still show up to conferences and meet in plenary or online, much of the interpersonal work of transnational norm-building and meaning-making in these venues has always occurred in informal social settings—conference venue hallways and lobbies, coffeeshops and after-parties—whose nature has been altered or affected by social distancing requirements and lockdowns. Another issue distinguishes the killer robot ban from earlier campaigns. Whereas landmines and clusters campaigners simply had to demonstrate that the humanitarian costs of landmines outweighed their military benefits, the humanitarian impacts of autonomous weapons systems are more speculative. Some weapons have been banned preemptively in the past—blinding lasers and flattening bullets, for example—but in these cases, much of the impetus came from an unlikely state leader with clout and a horse in the race: Russia and, during the Cold War, the USSR. The killer robot issue faces the pre-emption problem, but without great power leadership. Indeed, thus far several major states seem arrayed against the idea. Third, as political scientists Elvira Rosert and Frank Sauer write, the issue of exactly what is wrong with killer robots is open to framing in multiple ways. Should the argument be that they might harm civilians, for example, or that humans have a fundamental right not to be killed by machines acting outside of human control? Do campaigners want to ban all autonomous weapons systems or only the direct targeting of humans by algorithms? These questions have bedeviled diplomats and campaigners. Disputes among activists themselves about objectives and moral logic have been trickier to resolve on this campaign than for previous ones. That makes it easier for laggard states to muddy the framings even more and drive wedges within the movement. But campaigners are refining their message, while vaccinations are rendering transnational social spaces more open. And with advances in military technology adding urgency to the issue, time is on the campaigner’s side—if not humanity’s. The fact that campaigners have persisted and strengthened their momentum despite challenges and a global pandemic underscores the depth of their desire, and the desire of large swaths of the public, to ban the killing of humans by autonomous machines. Though the CCW failure is a boon to prospects for building a strong norm elsewhere, one more ingredient is now required in the humanitarian disarmament recipe: a state actor not only sympathetic to the idea of a killer robot treaty but seized enough—and with deep enough pockets—to take the lead on hosting and sustaining a separate multilateral process outside the U.N. orbit. The failure of the CCW process has created an opening for countries in favor of new international law on the issue—including New Zealand, Austria, Germany and others—to flex their humanitarian muscles, as Canada did for landmines and Norway and Ireland did for cluster munitions. Political scientist Naomi Egel’s new work shows that humanitarian disarmament advocacy offers middle power states opportunities to expand their clout and national brand by exercising moral leadership on the world stage. Who will capture the lead role in exercising convening power in this round of treaty-making remains to be seen, but it is likely to occur sooner rather than later. Two things seem certain: Governments won’t wait another five years to meet again about a killer robot treaty, and the next time they do, it won’t be in Geneva.

Charli Carpenter is a professor of political science and legal studies at University of Massachusetts-Amherst, specializing in human security and international law. She tweets at @charlicarpenter. Her weekly WPR column appears every other Friday.