Published: October 26, 2026
International humanitarian law (IHL) takes an “effect-based” approach to regulating/banning Lethal Autonomous Weapon Systems (LAWS). This “outcome-driven,” or result-oriented, approach is based on real-world results of a weapon system’s deployment. It suggests that a weapon system’s regulation or ban is justified only if it has been used or is proven to cause “excessive injuries” to human life. Because of this inherently outcome-driven approach, many legal frameworks struggle to address LAWS, whose opaque machine-learning algorithms make outcomes highly unpredictable. In contrast, Pakistan takes a “functional approach” to governing LAWS, based on their algorithmic design and inherent capabilities before they are ever used on the battlefield. In this regard, Pakistan’s Working Paper submitted to the Group of Governmental Experts (GGE) on the Convention on Certain Conventional Weapons (CCW) review conference in 2023, and the Proposal on Elements of an International Legal Instrument, submitted in 2024, contribute to the broader discussion on LAWS. These proposals primarily address the “effect-based gap” through “design-centric prohibitions” and regulations, given the lack of effective human control.
The core difference between the two approaches is clear in how they address the “use” or “pre-use” phases of lethal autonomous systems. The traditional effect-based approach relies on quantifying real-world outcomes, payload destruction patterns, and “visible consequences” as the primary trigger for regulation. It struggles with “quantifying” AI autonomy, as black-box algorithms make battlefield effects impossible to quantify accurately. The traditional approach has a “threshold of lethality” that is qualified only if the force deployed results in “actual lethal effects” or “illegal suffering.” Therefore, the traditional legal philosophy is more reactive and relies on retrospective “post-use” accountability.
Pakistan’s functional approach is based on inherent system design and autonomous capacity to select/engage targets. Regarding the autonomy of the system, unlike the traditional approach, Pakistan classifies a weapon as a LAWS if it operates without human intervention “after activation,” regardless of the outcome. Similarly, the lethality of the system is qualified if the system possesses the capability to apply lethal force by design, not necessarily yielding harmful outcomes, even if a specific strike “fails” or “misfires.” This proactive approach implements design-level and preemptive engineering prohibitions. However, despite its comprehensiveness and logical coherence, this ambitious stance may face opposition from parties invested in developing these systems, particularly for “non-lethal” and dual-use applications of LAWS. The “permissive camp,” as well as the “obstructers,” oppose such an approach beyond discussion on the grounds of dual-use blind spots, blurred thresholds of autonomy, software-hardware disconnect, and challenges of international verification and monitoring.
An appraisal of key characteristics of Pakistan’s working paper on a legal instrument for LAWS suggests that it effectively addresses and emphasizes state responsibility through a structured, “multi-tier” legal framework. The multi-pronged approach advocates an immediate global prohibition on systems that make life-and-death decisions about the use of force entirely outside human control. This is specifically important for systems that are incapable of distinguishing between active combatants and protected civilians and the LAWS that produce battlefield effects that cannot be adequately predicted, understood, or explained by human commanders. This classification is more in line with the tiered restrictions advocated by humanitarian organizations, which call for systems that do not meet the strict threshold for an outright ban. They advocate for legally binding operational boundaries rather than a voluntary code of conduct. Furthermore, they suggest that voluntary compliance frameworks or political declarations offer no true accountability and leave massive “compliance gaps” that are susceptible to exploitation by malicious actors.
Pakistan’s May 2024 submission, building upon its 2023 Working Paper, formalizes these concepts into coherent legislative text designed for becoming a binding international treaty governing LAWS. The key characteristics of these proposals suggest that the functional framework be negotiated and formally adopted as Protocol IV of CCW. It refines the scope of the legal instrument that must regulate the general category of LAWS by focusing on how they function rather than on the ever-shifting physical and technological characteristics of a particular system. The proposal heavily relies on codifying human-in-the-loop, which legally ties the concept of lethality to the activation phase rather than to the system’s outcome. It insists rightly that if a system selects and fires upon a target post-activation without explicit human confirmation, it violates the proposed protocol regardless of whether the “target survives.”
Pakistan’s proposal can shape the ongoing discussion at the CCW’s 2026 discussion cycle, within its critical three-year mandate to develop a legally binding instrument on LAWS. At this stage, as the GGE negotiates a “rolling text” to establish a standardized global baseline, Pakistan’s clear-cut functional definition provides a pragmatic bridge between states demanding a total ban and those pushing for permissive, voluntary regulations. The functional approach may help address the predictability standard as advances in machine learning suggest these systems may unpredictably change mid-deployment, rendering standard pre-deployment legal reviews obsolete (Article 36 reviews). The explainable and predictable characteristics of the functional approach bring the baseline cognitive ability of a human commander to the center of debate to understand weapons actions before activation.
In sum, Pakistan’s persistent and structured legal drafts have already unified a substantial voting bloc of developing countries that reject weak confidence-building measures. It may facilitate the collective demands that the 2026 final session deliver a binding instrument. The pragmatic approach is to impose a legally binding prohibition in areas where there is no effective human control, and to regulate the applications of LAWS where there is some human control, with the least uncertainty and unpredictability. Therefore, a functional approach to governing LAWS can play a significant role, at least in a rolling text comprising some compromise language beyond vague political statements and technical definitions.
Atta Ullah is a Research Fellow at Balochistan Think Tank Network (BTTN), Quetta. He focuses on Core issues at the Conference on Disarmament (CD) and on broader subjects of arms control, non-proliferation, and disarmament. The views expressed in this article are the author’s own.
