4 Scenarios for the Future of AI in Warfare

The integration of artificial intelligence into military operations is shifting global warfare from human-directed combat to algorithm-driven engagements, sparking a high-stakes arms race among major powers. While autonomous systems promise unprecedented operational speed and reduced human casualties, they simultaneously introduce profound risks of unintended escalation and uncontrollable drone swarms. The future of global security over the next decade will be largely determined by whether the international community can establish enforceable frameworks for meaningful human control before the technology completely outpaces diplomacy.

What Are Lethal Autonomous Weapon Systems?

When the general public thinks of artificial intelligence in warfare, the imagery is often dominated by science fiction narratives. Pop culture touchstones like the Terminator franchise conjure fears of sentient, humanoid machines that develop free will, rebel against their creators, and launch nuclear holocausts ¹¹. Defense analysts and technologists routinely caution that these theatrical analogies are not only scientifically inaccurate, but they also distract from the very real, immediate challenges posed by modern military AI ¹².

In reality, Lethal Autonomous Weapon Systems (LAWS) are machines that, once activated, can independently select and engage targets without further intervention by a human operator ³⁴. They do not possess consciousness, moral reasoning, or an independent desire to kill. Instead, they rely on complex algorithmic targeting, sensor data ingestion, and advanced machine learning models to execute highly specific, pre-programmed parameters ².

To grasp how this technology functions on a battlefield, it is far more accurate to look at everyday civilian technologies. Just as a commercial airliner's autopilot manages complex flight paths by continuously reading environmental sensors, or a smartphone utilizes computer vision for facial recognition, military AI applies these exact same principles to the theater of war ⁵⁶. Artificial intelligence allows an unmanned aerial vehicle (UAV) to navigate contested environments where GPS signals are jammed, using visual terrain-matching to find its way . It uses algorithmic computer vision to distinguish between a civilian pickup truck and a military rocket launcher based on massive datasets of training imagery ⁷.

The Evolution from Defensive to Offensive Autonomy

It is a common misconception that autonomous weapons are a completely novel phenomenon. Certain defensive systems incorporating rudimentary autonomous functions have been in active service for decades ⁸. For example, anti-vehicle and anti-personnel mines operate autonomously based on physical trigger mechanisms. More advanced automated sentry guns and naval missile defense systems - such as the Phalanx close-in weapon system - are programmed to automatically detect, track, and destroy incoming projectiles faster than a human operator could react ⁹.

The critical threshold currently being crossed is the leap from stationary, defensive automation to mobile, offensive autonomy ¹⁰. Modern developments focus on systems like loitering munitions, often referred to as "kamikaze drones," which can be launched into a designated geographic box to independently hunt for specific targets using onboard AI . By processing massive amounts of intelligence, surveillance, and reconnaissance (ISR) data at the "tactical edge" without needing a continuous data-link back to a human commander, these weapons act as force multipliers ⁶¹⁴.

Defining Meaningful Human Control

The debate over these weapons centers on the legal and ethical definition of "meaningful human control" or "appropriate levels of human judgment" ⁴¹¹. The International Committee of the Red Cross (ICRC) stipulates that a system is a fully autonomous weapon if it possesses autonomy in its "critical functions" - specifically the searching for, detecting, identifying, tracking, and attacking of targets ⁹¹².

Autonomy is generally categorized into three tiers of human involvement: 1. Human-in-the-loop (Semi-autonomous): The machine can track and suggest targets, but a human must explicitly authorize the firing of the weapon. 2. Human-on-the-loop (Human-supervised): The machine can select and engage targets on its own, but a human operator monitors the process and has the ability to abort or override the attack. 3. Human-out-of-the-loop (Fully autonomous): The machine is activated and completes the entire "kill chain" on its own, with no capacity for a human to intervene once the system is deployed ³¹¹.

Advocacy groups and prohibitionists argue that "human-out-of-the-loop" systems violate International Humanitarian Law (IHL), as algorithms cannot adequately apply the subjective legal principles of proportionality (weighing military advantage against civilian harm) and distinction (differentiating combatants from non-combatants) ¹²¹⁷. Conversely, military strategists counter that enforcing a strict "tight human-in-the-loop" requirement is rapidly becoming a fatal tactical liability ¹¹⁸. If an adversary deploys a swarm of hundreds of autonomous drones or hypersonic glide vehicles, human cognitive reaction times are simply too slow to manage the defense ¹¹⁹.

How Major Powers Are Adopting Military AI

The global market for autonomous weapons reached $14.2 billion in 2024 and is projected to nearly double to $27.6 billion by 2034, driven by a compound annual growth rate of over 21 percent ³¹⁴. This explosive growth is underpinned by multi-billion-dollar defense modernization programs across the globe. However, there is no monolithic approach to military AI. The world's major military powers are adopting distinctly different doctrines regarding how artificial intelligence should be integrated into their command structures and battlefields.

Comparing National Military AI Doctrines

The United States: Algorithmic Teaming and Rapid Swarms

The United States Department of Defense has long been a pioneer in automated systems, but it is currently undergoing a rapid doctrinal shift to counter peer adversaries. A persistent myth in the defense community is that U.S. policy explicitly prohibits fully autonomous weapons. In reality, DoD Directive 3000.09 - updated in 2023 - does not ban autonomous lethal targeting, nor does it strictly require a human-in-the-loop at the tactical level ⁴. Instead, it mandates a rigorous, senior-level review process to ensure that autonomous systems function safely, predictably, and in accordance with U.S. law before they are fielded ⁴.

The U.S. strategy heavily relies on the commercial tech sector to scale up capabilities quickly. For example, the Pentagon's Replicator initiative aims to field thousands of autonomous systems across multiple domains to counter the mass of adversary forces. More recently, in early 2026, the U.S. pushed forward Project LUCAS (Low-Cost Uncrewed Combat Attack System) ²⁰. By integrating advanced autonomy stacks - like Shield AI's "Hivemind" - onto reverse-engineered, commodity airframes, the U.S. aims to field swarms of attack drones that cost roughly $35,000 each, compared to the $2.5 million price tag of a traditional Tomahawk missile ²⁰. This economic asymmetry allows a single human operator to orchestrate hundreds of autonomous drones simultaneously, fundamentally shifting the math of military deterrence.

China: The Transition to "Intelligentized" Warfare

The People's Liberation Army (PLA) of China has officially shifted its modernization goals away from merely "informatized" warfare toward what it calls "Intelligentized Warfare" ¹⁴²⁰³⁰. This doctrine, formalized in defense white papers, posits that the future of combat relies on the seamless integration of AI, robotics, quantum computing, and big data to secure a decisive, systemic advantage ¹⁴³⁰. The goal is to create a networked "system of systems" by 2027 - a milestone tied to the centennial of the PLA - that provides Beijing with credible military options, particularly regarding a Taiwan contingency ²¹.

China is heavily investing in converting commercial innovation into combat power through its military-civil fusion strategy. This includes the development of robotic quadrupeds (referred to as "robotic wolves"), AI-powered electronic warfare (EW) systems capable of suppressing radar signals from thousands of miles away, and autonomous drone carriers ¹⁴²⁰.

Paradoxically, while racing to deploy military AI, China has also been highly active in shaping global AI governance diplomacy. Through its Global AI Governance Initiative, launched by President Xi Jinping in 2023, China advocates for a "people-centered approach" to military AI, explicitly calling on nations to refrain from seeking absolute military advantage through AI ¹³³². In November 2024, China and the U.S. achieved a landmark bilateral agreement in Lima, affirming that human beings, not AI, must maintain control over nuclear weapons decisions ³². Observers note that China's diplomatic efforts serve a dual purpose: positioning Beijing as a responsible global steward while attempting to shape international standards that serve its own strategic preferences ³²²².

Russia: Applied Autonomy and Nuclear Thresholds

Russia's approach to military AI is highly centralized, deeply pragmatic, and increasingly tied to its broader survival in the context of intense international sanctions and the ongoing war in Ukraine. Rather than competing with the U.S. or China in foundational artificial general intelligence, Russia focuses on "applied AI" - adapting existing open-source models to integrate machine learning directly into tactical combat systems ¹⁵.

Faced with massive attrition in traditional armored forces, Russia has pivoted to building a "sovereign drone ecosystem." The state has set aggressive targets, aiming to produce 130,000 large-scale unmanned aircraft systems annually and mandate the training of over 15,000 AI specialists per year by 2030 ¹⁵. A critical component of this modernization is the "Svod" system, an AI-enabled command-and-control architecture designed to aggregate video feeds, operator telemetry, and strike data to augment human situational awareness ²³. On the battlefield, Russia is increasingly deploying platforms capable of navigating and selecting targets without GPS or external communication, providing functional independence in environments where electronic warfare is intense ¹⁵.

Crucially, Russia's embrace of AI coincides with significant shifts in its broader military posture. In November 2024, President Vladimir Putin signed a decree updating Russia's nuclear doctrine. The update explicitly lowered the threshold for nuclear weapons use, authorizing a nuclear response not only to existential threats but also to conventional attacks or massive drone swarms that create a "critical threat" to the sovereignty or territorial integrity of Russia and its ally, Belarus ²⁴²⁵. At the United Nations, Russia has consistently pushed back against efforts to ban autonomous weapons, arguing that current international humanitarian law is sufficient and explicitly rejecting terms like "meaningful human control" as subjective ¹²²⁶.

Israel: The AI and Autonomy Administration

Israel has long been a global hub for technological innovation, and its military is systematically institutionalizing artificial intelligence. On January 1, 2025, the Israeli Ministry of Defense officially inaugurated the "AI and Autonomy Administration" ¹⁶²⁷. Operating under the Directorate of Defense Research & Development (DDR&D), this new administrative body was established to centralize and accelerate the deployment of autonomous systems across all branches of the Israel Defense Forces (IDF) ¹⁶¹⁷.

The creation of the AI and Autonomy Administration aligns with the IDF's "Hoshen" multiyear plan (2026-2030), which prioritizes robotics and AI to build a more lethal, casualty-resistant force ²⁴. Israeli defense leaders, including Ministry Director General Eyal Zamir, envision a future battlefield dominated by mixed human-machine units, where unmanned ground vehicles handle high-threat reconnaissance and clearance tasks, while human soldiers operate safely from supervisory roles ²⁴¹⁶.

Israel's aggressive pursuit of military AI is heavily informed by its recent operational experiences. During ongoing conflicts in Gaza and Lebanon, the IDF has utilized AI-based decision-support systems, such as "Lavender" and "Habsora" (The Gospel), to rapidly generate target recommendations based on vast intelligence datasets ³³⁹. While Israeli officials maintain that human analysts review all AI-generated recommendations to ensure legal compliance, the unprecedented speed and volume of algorithmic targeting have sparked intense international scrutiny regarding the ethical application of AI in dense urban warfare ³³⁹.

Iran: Asymmetric Swarms and Proxy Networks

Heavily constrained by decades of Western sanctions, Iran cannot compete directly with the U.S. or China in the development of frontier AI models ¹⁸. Instead, Tehran employs an asymmetric strategy, leveraging low-cost, AI-enhanced technologies to project power and coerce regional adversaries.

Iran has established a formidable domestic drone industry, highlighted by the Shahed-136 loitering munition. By integrating basic machine learning and autonomous flight capabilities into cheap, mass-produced airframes, Iran can launch swarm attacks designed to saturate and exhaust the expensive air defense systems of its rivals ²⁰¹⁸. This capability is not kept strictly for domestic use; Iran routinely proliferates these weapons to proxy militias across Lebanon, Syria, Iraq, and Yemen, effectively exporting autonomous strike capabilities across the Middle East ¹⁸²⁸.

In January 2025, Iranian Defense Minister Aziz Nasirzadeh announced that the Iranian Army had received over 1,000 new "strategic, stealth, and anti-fortification" drones equipped with artificial intelligence ¹⁹. This marked a notable shift in Iran's defense doctrine, signaling a heavy investment in the conventional Army rather than just the Islamic Revolutionary Guard Corps (IRGC) ¹⁹. These asymmetric capabilities were put to the test shortly after; following a brief exchange of hostilities in October 2024, a more intense international armed conflict - dubbed Operation Rising Lion - erupted between Iran and Israel in June 2025, highlighting the growing lethality of AI-guided aerospace threats in the region ²⁹.

4 Scenarios for the Future of Military AI

As the global autonomous weapons market accelerates, the international community stands at a crossroads. Over a decade of deliberations at the United Nations Group of Governmental Experts (GGE) on lethal autonomous weapons has yielded limited progress, hampered by deep divisions over definitions and regulatory approaches ³⁰. With a crucial UN mandate decision looming in 2026, defense analysts, think tanks, and wargaming experts have modeled how the militarization of AI might unfold by 2030 ³⁰³¹⁴⁴.

Based on these strategic foresight programs, four distinct scenarios emerge, ranging from utopian global cooperation to catastrophic algorithmic conflict ⁴⁵⁴⁶.

Scenario 1: The Global Orchard (Regulated Consensus)

In the most optimistic scenario, the terrifying pace of military AI development acts as a catalyst for global unity, much like the shadow of nuclear annihilation spurred Cold War arms control treaties ³¹⁴⁵. In "The Global Orchard," geopolitical tensions cool sufficiently to allow world leaders to forge a binding international consensus on AI regulation.

In this future, the United Nations establishes a powerful International AI Agency - reminiscent of the International Atomic Energy Agency (IAEA) - tasked with enforcing ethical principles and technical standards across all member states ⁴⁵⁴⁷. A legally binding instrument is ratified that explicitly prohibits fully autonomous lethal weapon systems from operating without meaningful human control, effectively banning "killer robots" while allowing militaries to use AI for non-lethal logistics, cyber defense, and intelligence analysis ⁴⁵. Robust liability frameworks ensure compliance, fostering a boom in responsible, civilian-focused AI innovation while putting a hard ceiling on the militarization of autonomous combat ⁴⁵.

Scenario 2: Walled Gardens and the Digital Iron Curtain

Rather than achieving global consensus, the "Walled Gardens" scenario envisions a world fractured into heavily guarded, competing technological blocs ⁴⁵⁴⁸. In this reality, the United States and its allies build one AI ecosystem grounded in democratic norms and Western hardware, while a Sino-Russian bloc develops a parallel, sovereign AI infrastructure ⁴⁸.

This scenario is defined by a "Digital Iron Curtain." Supply chains completely decouple as nations realize that relying on foreign AI architecture is a severe national security vulnerability ⁴⁸. Export controls on advanced semiconductors become the primary weapons of this new Cold War, intended to choke off the computational power required for rivals to train frontier AI models ³⁰⁴⁸.

Within these walled gardens, military AI develops rapidly but unevenly. Because ethical standards and legal norms fail to align globally, each bloc aggressively fields autonomous systems designed specifically to exploit the digital architecture of its adversaries ⁴⁵⁴⁸. The arms race is managed and contained within these spheres of influence, leading to a tense, highly militarized stalemate where technology serves as the ultimate geopolitical leverage.

Scenario 3: Rapid Proliferation and Asymmetric Swarm Warfare

If technological diffusion outpaces both regulation and export controls, the world enters a phase of dangerous proliferation ¹⁴. The "Rapid Proliferation" scenario is driven by the plunging costs of commercial drone technology and the widespread availability of open-source artificial intelligence models ⁴⁸.

In this future, the monopoly on advanced warfare is broken. Middle powers, authoritarian regimes, and non-state actors easily acquire or reverse-engineer sophisticated autonomous capabilities ³¹. The nature of conflict shifts dramatically as traditional, multi-billion-dollar military assets - like aircraft carriers or advanced fighter jets - become highly vulnerable to swarms of thousands of cheap, AI-driven suicide drones ¹⁹.

Because these autonomous systems rely on edge computing rather than continuous communications links, they are highly resilient against traditional electronic warfare and GPS jamming ¹⁴. The widespread availability of lethal autonomy makes warfare cheaper, more frequent, and exponentially harder to attribute, leading to persistent, localized conflicts across the globe where defensive systems are constantly overwhelmed by asymmetric swarm tactics.

Scenario 4: Algorithmic Escalation and Loss of Control

The most severe scenario - "Algorithmic Escalation" - materializes if militaries fully embrace "human-out-of-the-loop" autonomy to prioritize speed above all else ¹. In this dystopia, the integration of algorithmic targeting without human verification leads to profound instability.

When two opposing, fully autonomous military systems interact in a contested environment, the algorithms can react to one another at machine speed. A minor sensor error or a spoofed signal could trigger an immediate, lethal response from one autonomous system, instantly provoking a counter-response from the adversary's AI. Because humans are removed from the decision cycle, this loop can spiral into a catastrophic "flash war" in a matter of seconds, long before political leaders are even aware a crisis has begun ¹.

Furthermore, extreme reliance on autonomous systems introduces severe internal security risks. Researchers have theorized the threat of "AI-enabled coups." If a state's military architecture is entirely automated, a rogue faction or high-ranking official could program millions of autonomous weapons to be singularly loyal to themselves ⁴⁹. By bypassing the traditional human chain of command - which historically requires convincing thousands of soldiers to betray their government - a small group with exclusive access to the AI's core code could seize control of a nation with terrifying efficiency ⁴⁹.

Bottom line

The weaponization of artificial intelligence is no longer a theoretical debate reserved for science fiction; it is an active operational reality that is currently reshaping battlefields and defense budgets worldwide. From the United States' pursuit of low-cost autonomous swarms to China's vision of intelligentized warfare and Israel's dedicated AI military administration, the advantage in future conflicts will unequivocally belong to the forces that best integrate algorithmic speed. As the international community faces critical UN deadlines in 2026, the ultimate challenge is forging a consensus on meaningful human control before the proliferation of autonomous weapons leads to uncontrollable algorithmic escalation.