Autonomous Weapons vs Manned-Unmanned Teaming

The Human Control Spectrum

The most fundamental difference between autonomous weapons and manned-unmanned teaming is where each sits on the spectrum of human control. Autonomous weapons operate at the far end—systems like AI-guided kamikaze drones in Ukraine lock onto targets identified during terminal flight, with AI replacing what CSIS estimates is 99% of the human labor in the engagement chain. At the extreme, fully autonomous systems complete the entire find-fix-finish cycle without any human input.

Manned-unmanned teaming deliberately preserves the human at the center. The pilot functions as a mission commander rather than a direct controller, issuing intent-level commands while AI handles tactical execution. This mirrors the multi-agent systems paradigm in civilian AI, where an orchestrating agent delegates tasks to specialized sub-agents. The U.S. Navy's January 2026 tests demonstrated this model: BQM-177A targets flew autonomously using Shield AI's Hivemind software while a virtual F/A-18 acted as mission lead, proving the human-in-command concept at operational fidelity.

This distinction has profound legal implications. MUT architectures generally satisfy the "meaningful human control" standard demanded by international humanitarian law, while fully autonomous weapons systems face an unresolved governance vacuum that 156 nations voted to address in the November 2025 UN General Assembly resolution.

Battlefield Economics and Force Design

Autonomous weapons and MUT solve different economic problems in force design. Autonomous weapons—particularly the expendable drone swarms proliferating in Ukraine—exploit radical cost asymmetry. A $500 first-person-view drone destroying a $5 million armored vehicle represents the kind of exchange ratio that reshapes warfare. Ukraine scaled drone production from 2.2 million units in 2024 to 4.5 million in 2025, with projections of seven million in 2026, demonstrating industrial-scale attrition warfare.

MUT addresses a different equation: how to extend the combat power of expensive crewed platforms without proportional cost increases. The U.S. Air Force's CCA program, funded at $8.9 billion through 2029, aims to pair $2–4 million loyal wingmen with $80+ million F-35s. A fleet of ten autonomous wingmen plus one crewed fighter delivers far more capability per dollar than two crewed fighters. But these are reusable, sophisticated platforms—not expendable munitions.

The strategic implication is that autonomous weapons democratize lethality (any nation or non-state actor can field kamikaze drones), while MUT reinforces the advantage of technologically advanced air forces. For the United States, MUT is the preferred paradigm because it preserves the qualitative edge of its crewed fleet. For actors without advanced aviation, autonomous weapons offer an asymmetric path to military capability.

AI Technical Demands

The AI challenges in each domain are qualitatively different. Autonomous weapons require reliable computer vision for target identification, terminal guidance algorithms, and increasingly, swarm intelligence for coordinated multi-drone operations. The AI must be good enough to distinguish valid targets from civilians and friendly forces—a problem that remains unsolved at the reliability levels required by international humanitarian law.

MUT demands a more complex AI stack. Loyal wingmen must process sensor data, maintain formation at supersonic speeds, avoid threats, execute tactical maneuvers, and respond to degraded communications—all with robust onboard edge inference rather than cloud-dependent processing. This is embodied AI operating in the most adversarial conditions imaginable: electronic warfare environments where GPS and data links may be jammed or spoofed. The October 2024 incident where Russia shot down its own S-70 Okhotnik-B after losing control illustrates the lethal consequences of AI failure in this domain.

Both domains push the frontier of reinforcement learning for tactical decision-making, but MUT adds the multi-agent coordination challenge—multiple autonomous platforms must collaborate in real time while maintaining alignment with human commander intent.

The Ukraine Laboratory

The war in Ukraine has become the defining proving ground for autonomous weapons, accelerating development timelines by years. AI-guided drones now account for 70–80% of battlefield casualties. Ukraine's innovation cycle—from improvised commercial drones to AI-enabled swarms with autonomous target tracking—has compressed what would normally be decades of military R&D into months. In March 2026, Ukraine took the unprecedented step of opening its battlefield AI data to allied nations and defense companies, creating a shared training dataset for autonomous systems drawn from real combat.

MUT, by contrast, has not yet been tested in high-intensity combat. Its development remains in the structured acquisition pipeline of advanced militaries—flight testing, integration exercises, and planned initial operating capability in the late 2020s. The December 2025 MQ-28 Ghost Bat missile engagement was a landmark achievement, but it occurred in a controlled test environment, not under fire. This gap between the battle-hardened reality of autonomous weapons and the theoretical promise of MUT is one of the most important asymmetries in the current defense landscape.

International Governance and the 2026 Crossroads

The governance trajectories of these two concepts could not be more different. Autonomous weapons face a potential regulatory reckoning at the 2026 Seventh Review Conference of the Convention on Certain Conventional Weapons in Geneva. The Group of Governmental Experts on LAWS met in March 2026 and will reconvene in August–September 2026, working toward what many observers consider the last realistic window for a binding international agreement. If no treaty emerges, the pace of autonomous weapons development—driven by major powers that have resisted restrictions—will likely make future regulation obsolete before it can be implemented.

MUT faces no comparable governance challenge. Because it retains meaningful human control by design, it operates comfortably within existing international humanitarian law frameworks. This regulatory clarity is itself a strategic advantage: nations can invest in MUT programs without the legal uncertainty that shadows autonomous weapons development. The U.S. Department of Defense's updated directive on autonomy in weapons systems explicitly supports human-supervised autonomous operations—the exact model MUT embodies.

The Convergence Ahead

Despite their differences, autonomous weapons and MUT are converging. Future MUT architectures will likely include expendable autonomous munitions launched from loyal wingmen—combining the human command authority of MUT with the speed and scale of autonomous engagement. The CCA program already envisions wingmen that can switch between tightly supervised modes (where the human approves every action) and more autonomous modes (where the CCA reacts to threats faster than a human could direct). China's Feihong FH-97 and the U.S. Replicator initiative both point toward hybrid force designs that blend crewed command, autonomous wingmen, and expendable autonomous munitions into integrated swarm architectures.

The question is not which approach will win, but how they will merge—and whether international norms can keep pace with the resulting capabilities. The decisions made in 2026, both in Geneva's conference rooms and on Ukraine's battlefields, will shape that answer for decades.