Myths vs. Fact: Autonomous Weapons and the Global Effort to Control Them

Autonomous weapons are by no means a concern for the distant future. Not only do they play a significant role in ongoing conflicts, but their use is rapidly increasing. While it can be challenging to identify the exact moments when these weapons are used, there is significant evidence of their widespread presence in recent conflicts. Most notably, the war in Ukraine is seen as the first large-scale conflict involving the widespread use of drones by both sides. As reported in Wired, there is growing pressure to “hand over more and more control to artificial intelligence,” potentially leading to systems that can operate on the battlefield with minimal human involvement.

For more details, visit Autonomous Weapons Watch, an organization dedicated to tracking the use of autonomy and AI in weapon systems.

Those opposing international regulations on autonomous weapons often argue that given their alleged ability to make precise, data-driven decisions without human error, autonomous weapons will reduce collateral damage, resulting in fewer civilian casualties. Human soldiers operate in high-pressure, high-stress environments, often with little sleep, making them highly susceptible to mistakes. By contrast, autonomous weapons are not affected by factors like stress, fear, or fatigue. Moreover, because they lack a self-preservation instinct (unless specifically programmed), these systems could theoretically eliminate instances where innocent civilians are killed by nervous soldiers acting with a “shoot-first, ask questions later” mentality.

While the promise of autonomous weapons to reduce collateral damage and civilian casualties may seem appealing, the reality is far more complex. Autonomous weapons rely on algorithms that, though advertised as precise, may not account for the nuances of human judgment, especially in rapidly evolving environments. Additionally, these systems are vulnerable to programming errors, biased data, and misinterpretations of battlefield conditions, all of which could lead to the misidentification of targets and unintended civilian casualties. Without human oversight, the risk of accountability for mistakes diminishes, increasing the likelihood that autonomous weapons could violate international humanitarian law and inadvertently escalate harm rather than reduce it.

There is also concern that the speed and scale at which autonomous weapons can be deployed will significantly amplify their potential for mass destruction. A single operator could unleash widespread devastation, with little to no control over the rapid escalation of force. In wargame models, the American think tank RAND found that “the rapid pace of autonomous systems contributed to inadvertent escalation,” underscoring the heightened risks associated with their deployment. Even if autonomous weapons occasionally make more ethical decisions than humans, the overall risk of conflict is significantly amplified by the speed at which they can act, raising concerns about the broader consequences of their use in warfare.

Finally, while autonomous weapons may become more skilled at distinguishing between human targets as they advance, these same capabilities could also enhance their potential for committing atrocities, such as genocide and targeted killings. This raises significant ethical and legal concerns, as increased precision does not guarantee compliance with humanitarian principles.

Due to differing interpretations of what constitutes a truly autonomous weapons system, there is currently no unanimous definition. Israel’s HARPY drone, for example, autonomously seeks out and destroys radar systems, yet requires human operators to pre-program the targets and launch the drone. While reaching a consensus on a clear definition would be beneficial for creating effective regulations, it is not a prerequisite for action. Often, this lack of agreement is used as a red herring by states opposed to a legally binding instrument. In previous arms control treaties, such as those on cluster munitions and biological weapons, significant progress was made before establishing precise definitions. This demonstrates that meaningful regulation can be achieved without waiting for a perfect definition.

In November 2013, parties to the UN Convention on Certain Conventional Weapons (CCW) agreed on a new mandate to initiate discussions on the development of specific, legally binding regulations for the development, deployment, and use of autonomous weapons systems. It is often argued that this is sufficient to address the risks posed by autonomous weapons and that we should allow more time for the CCW to fulfill its mandate.

The reality, however, is that the CCW has significant limitations that hinder its global effectiveness. It excludes 66 member countries, restricting its reach and diminishing the inclusivity of its discussions. Furthermore, the interpretation of its scope by several states is confined to the conduct of war between states under international humanitarian law, neglecting critical concerns such as ethical implications, human rights violations, and the proliferation of weapons to non-state actors. Additionally, the CCW does not address how autonomous weapons could facilitate future atrocities, such as genocides. Despite some progress in building common understandings, advancements have often been stalled by states exploiting the CCW’s consensus-based decision-making process, which requires unanimous agreement on all decisions. As such, the CCW’s role as the primary forum for addressing these issues is increasingly in question.

After the Sixth Review Conference in 2021 failed to reach a consensus on establishing binding controls, states began exploring regional conferences and the 2023 General Assembly resolution in an effort to break the persistent deadlock. Given that the CCW has been discussing autonomous weapons for over a decade without significant progress, it is reasonable to question whether it can ever truly fulfill its mandate in its current form.

The CCW has not produced any new international law since 2003. 

Further information on the CCW can be found in our post Key Forums Shaping the Global Debate on Autonomous Weapons Policies.

Geopolitical tensions are undoubtedly high in many parts of the world today, with ongoing conflicts and rivalries straining international relations, such as the Russia-Ukraine war, which has exacerbated tensions between Russia and the West to levels not seen since the Cold War. The notion that this presents an insurmountable barrier to meaningful progress on arms control is contradicted by history. Geopolitical tensions have always existed, often even more intensely than they are today, yet this has not prevented meaningful progress on critical international issues.

During the Cold War, the world was divided by an ideological and military standoff between the US and the Soviet Union. Nevertheless, international agreements such as the Strategic Arms Limitation Talks (SALT) and the Nuclear Non-Proliferation Treaty (NPT) were successfully negotiated. Bioweapons offer another example. Once viewed as an essential military tool, the 1972 Biological Weapons Convention (BWC) effectively banned the biological and toxin weapons, prohibiting their development, production, acquisition, transfer, stockpiling and use.

These agreements helped prevent the escalation of global tensions and laid the groundwork for future diplomacy. 

The notion that international treaties lack influence when key countries like Russia, the US, or China are not part of them is another misconception easily disproved through historical evidence. Treaties can still have a profound impact on limiting the proliferation of weapons and shaping global norms, even in the absence of major powers. A notable example is the 1997 Landmine Treaty, which the US, Russia, and China did not ratify. (Though the US would later join the Ottawa Process and disavow the use of landmines, if only temporary) Despite their absence, the treaty increased the stigmatization of landmines, compelling numerous nations—both signatories and non-signatories—to implement national bans and drastically reduce their use.

Moreover, the risks posed by non-state armed groups further underscore the potential for such treaties to foster regional stability, even if key global powers do not sign on initially. Unlike nuclear weapons, which are controlled by a select few states, autonomous weapons are more accessible to a wider range of actors, making them more susceptible to regulation through well-crafted international agreements. A treaty that establishes a global norm, supported by a majority of nations, could significantly reduce the spread of these technologies. This could be reinforced by informal agreements between powers like the US and China, committing to abide by core regulatory principles.

Countries like Germany and the Netherlands, both of which are involved in the development and use of autonomous weapons, have already expressed support for negotiating legally binding rules (See our article The Political Landscape: Global Perspectives on Regulating Autonomous Weapons for more information). This growing momentum among states reinforces the idea that even in the absence of certain major powers, international regulation can still exert a powerful regulatory effect.