How the UN Bought Time for the World to Understand Autonomous Weapons

As the UN Group of Governmental Experts (GGE) concludes its decade-long mandate on lethal autonomous weapon systems (LAWS), its final report does little more than reiterate existing International Humanitarian Law (IHL). However, its value should not be judged as a failed legal endeavour, but as a protracted holding pattern that absorbed public moral panic and protected against premature vetoes or restrictions. In delaying top-down regulation, it bought essential time for States and armed forces to reveal the principles and practices of AI-enabled LAWS. This process provided opponents and proponents with technical ground truth rather than sci-fi hype.

Moral Panic and a Decade of Dither

The humanitarian risks from “Killer Robots” digitally dehumanising combatants and killing civilians dominated the public conscience in the early 2010s. The combination of uncrewed aircraft and emerging AI presented a fantastical science fiction threat that urgently needed to be controlled or banned. The GGE was established in 2017 to respond to concerns that existing controls were insufficient for novel LAWS.

As a mechanism to inform global regulation, the GGE’s initial progress appeared painfully slow, with huge effort expended just trying to define the problems it was meant to solve. In 2019, it agreed on eleven “guiding principles” to shape the group’s work. In 2023 the UN Secretary General called on world leaders to launch negotiations of a new legally binding instrument, with the GGE mandated to formulate the “set of elements” to guide this future work. As the 2026 Report concludes this task of international consensus-building and prepares the ground for even more negotiations, we must ask whether the GGE’s involvement was necessary or desirable?

Kicking LAWS Regulation into the Long Grass

The panic surrounding LAWS made political outsourcing inevitable. Orben’s The Sisyphean Cycle of Technology Panics highlights political distancing as an established and typical response to threats from new technology. The cycle’s first stage, “panic creation,” is characterised by actors highlighting the risks of new technology in broad, and often misleading, terms. For LAWS, this included the Stop Killer Robots Campaigners’ Kit, which helped create widespread fear of autonomous killing.

This moral panic then feeds the second stage, “political outsourcing,” whereby politicians push the debate to external contestation. LAWS contestation was outsourced to the GGE and its multinational, NGO and scientific membership.

Despite the GGE’s protracted contestation, its eventual consensus has produced very little that is different from the prohibitions and restrictions already in place for other weapons. The overwhelming majority of the report can therefore be seen as standard weapons controls.

In the report, IHL remains the central principle for LAWS use, and there are a few LAWS-specific elements to guide their design. For example, software must be locked down and understood, operational effects must be limitable, and a human must be responsible and accountable for each use. The requirement that “those responsible for the use of LAWS can limit their effects” is already a fundamental requirement of command and control.

Consensus that “IHL applies fully to the development and use of LAWS” and that “it is prohibited to make the civilian population … the object of an attack by LAWS” are just statements of the obvious. IHL applies to all weapons, and LAWS helps clarify this link by including “weapon” in its title (paras. 29, 33).

Real Progress Happened Elsewhere

Much progress has been made elsewhere since the GGE’s formation that make its efforts look lethargic and its outputs lightweight. The Responsible Artificial Intelligence in the Military Domain Summit (REAIM) reached a remarkably similar international consensus much quicker. In 2024, 58 States endorsed its normative framework, which establishes IHL as a foundational principle for LAWS use, requires human involvement as an essential condition, and sets a range of technical requirements. REAIM’s bottom-up approach focused on technical dialogue and avoided the UN’s politicised discourse.

A key output from REAIM was its invitation to States to develop national frameworks, strategies, and principles for the responsible use of military AI. In doing so, they revealed their autonomous futures.

The U.S. Department of War Directive 3000.09 Autonomy in Weapon Systems comprehensively sets out how the Department will mitigate the risks associated with LAWS. From initial concept, through testing, legal review, and to their deployment and use, it demonstrates careful consideration of autonomy and safety focused employment.

The UK’s Ambitious, Safe, Responsible AI Strategy and its Autonomous Collaborative Platform Strategy describe how its autonomous platforms will be used collaboratively and directed by humans. Australia demonstrated an MQ-28A Ghost Bat destroying a test target with an AIM-120 air-to-air missile, directed by human operators to fulfil a traditional air-to-air role. All these examples of operational design and implementation impatiently preceded, through necessity, the UN’s top-down direction.

LAWS Maturity Changed the Debate

These preparations and tests can now inform international regulation and demonstrate David Collingridge’s “dilemma of control.” He argues that regulating too early kills the technology before it is understood, whereas regulating too late may be meaningless. The dilemma is that, while early interventions into technological developments may achieve the strongest controls, it is necessary to delay regulation until the technology is mature enough to understand its implications. At that point, the technology may have become so embedded that regulation then becomes disruptive or even impossible. Timing is critical.

Activities outside the UN, where States independently developed their concepts of operation and candidate LAWS, revealed the regulation appropriate to real-world cases. This may represent Collingridge’s “tipping-point” moment, where the development of comprehensive protocols allowed States to demonstrate that a ban or significant restriction is unnecessary.

States’ LAWS aspirations are relatively modest: limited capability systems that collaborate with humans, featuring narrow AI, and tested and trusted to fulfil a limited number of specific tasks. Opponents and proponents now share a common understanding. The imagined free-roaming killing machines that fuelled the moral panic now appear fantastical science fiction compared to the mundane reality of paperwork and restrictive protocols.

The Revealed Function of the GGE

If States and their armed forces are willing and able to implement their autonomy plans before the UN GGE creates regulations for them, what is the point of the GGE? Is the process useful, or merely performative? Why do States participate in a consensus-driven approach where national agendas must be massaged into a single, but generic outcome?

One explanation is that GGE participation is a controlling move. States seek to keep dialogue trapped within the GGE (and subsequent groups) to avoid binding limits. By remaining in the discussion for consensus, States that oppose strict regulation retain a veto against it.

Another explanation is that GGE participation is precautionary. More than 70 States support a binding treaty so anti-treaty States, such as Russia and the United States, may choose to remain in the GGE rather than risk pro-ban coalitions making alternate arrangements. Ban options include taking negotiations to the UN General Assembly, or launching an independent process based on majority rules, such as the Ottawa Landmine Treaty. Continual dialogue denies these alternate routes.

“Political outsourcing” to the GGE also provided a pressure relief while the public adapted to the new future. Four years of drone use in Ukraine (albeit remotely piloted rather than AI-enabled autonomy) has shaped opposition to autonomy and desensitised the public to the use of drones. The moral panic is largely quelled despite reports of Ukraine planning to build 10 million drones in 2027.

A Performative but Useful Process

By adding nothing new to the existing IHL construct, the GGE’s final output is legally irrelevant. It is also practically redundant, arriving long after States developed practical LAWS solutions. There is nothing new that was not apparent a decade ago in alternative fora. Its real value was never regulatory but political.

The UN’s role in autonomous weapon regulation should be understood as a strategically functional performance rather than a failed legal endeavour. Although some participants may have considered urgent effort to be essential in avoiding an autonomous dystopia, the GGE’s structure provided a regulatory delay that proved useful for this immature, politically contentious, and rapidly evolving technology.

Politically, delegating contestation to Geneva absorbed public pressure, and lengthy discussions bought valuable time while the operational aspiration of safe military AI took shape. It placed autonomous weapon contestation in a performative holding pattern that avoided premature bans and overly restrictive controls. This is the “dilemma of control” in plain sight. It would have been easy to overreact to the moral panic and ban LAWS before their true nature was known. By delaying regulation until real evidence was available, this slow-burn performance therefore provided both “sides” with ground truth rather than sci-fi hype.

In November 2026, the GGE’s Review Conference will decide what happens next. Whether that is a definitive agreement, a mandate for negotiations, or another continuation of the process, the performance continues.

***

Michael Page is a lethal autonomous weapon systems researcher and leads a research programme focusing on their practical command and control. He holds an MPhil in Politics and International Studies from the University of Cambridge, and an MSc in Aerospace Systems.

The views expressed are those of the author, and do not necessarily reflect the official position of the United States Military Academy, Department of the Army, or Department of Defense.

Articles of War is a forum for professionals to share opinions and cultivate ideas. Articles of War does not screen articles to fit a particular editorial agenda, nor endorse or advocate material that is published. Authorship does not indicate affiliation with Articles of War, the Lieber Institute, or the United States Military Academy West Point.

Photo credit: U.S. Army photo by Spc. Josefina Garcia