A soldier who fires on a target must be able to reason “why” afterward. Similarly, a commander who authorizes a strike must be able to justify that decision to superiors, at times to lawyers, and possibly to the public. If an algorithm makes these decisions, how is the reasoning understood when it is not understood by the engineers who built it? Judging from how military technology has advanced, this is not hypothetical. It is the situation many militaries already face as artificial intelligence (AI) takes on a growing role in identifying threats, and even recommending or engaging targets. The usual conversation about military AI focuses on speed, power, and who can compute faster or strike first. But a quieter, more important question emerges alongside it. Which systems can justify their decisions when lives, laws, and legitimacy are on the line? Quantum AI, an unlikely and still unproven contender, is being explored as a possible answer. 

What Quantum AI Actually Is 

Quantum computing provides a fundamentally different approach to processing information than classical computer methods. Rather than relying solely on classical computational methods, quantum computers exploit quantum mechanical phenomena such as superposition, entanglement, and interference to solve certain classes of problems more efficiently than conventional computers. Where Artificial intelligence is the set of processes that let computers recognize patterns and make predictions with limited human guidance. Quantum AI combines the two, using quantum computers to conduct calculations while classical AI systems manage the rest of the process. 

Militaries are interested in this combination for several reasons, including faster logistics planning and better analysis of satellite imagery. Much of the focus on quantum AI has been on speed, optimization, and solving problems that challenge classical computers. But researchers are increasingly focused on a different possibility, one that has nothing to do with how fast a system can think and everything to do with how it decides and whether one can trust its decisions. 

The Problem With Black Boxes 

Most modern AI systems, particularly the neural network models that power image recognition and predictive analysis, function as what experts call a black box. The system takes in information and produces an answer, but its internal reasoning is buried in millions of numerical values that even its designers cannot fully interpret. For everyday uses like recommending a movie or predicting traffic, this opacity is harmless. In a military context, it is not. If an AI system flags an object in satellite imagery as a probable weapon launcher or recommends that a target meets the legal criteria for engagement, someone must eventually explain the decision reasoning. This is because militaries operate under rules of engagement and international humanitarian law, both of which require distinguishing combatants from civilians and justifying the use of force. If an investigation follows a strike, whether from a military tribunal, a human rights body, or a domestic court, “the algorithm decided” may not be an acceptable answer. Accountability requires a trail of reasoning and legal justification that a human being can follow and defend. 

This is not a distant, theoretical scenario. Militaries around the world are already integrating AI into surveillance, target prioritization, and threat classification systems. As these tools take on more responsibility, the gap between what they decide and what humans can explain grows wider. That gap creates legal exposure, operational risk, and serious human accountability problems for any military that wants to claim its use of force is lawful and proportionate. 

Why Quantum AI Might Offer a Way Out 

This is where quantum AI enters the conversation, not for its speed, but potentially to provide insight into AI decision making. Some researchers argue that certain quantum AI models, particularly a type known as gate-based models, organize information in ways that more closely mirror the actual structure of the problem being solved. Rather than compressing logic into an impenetrable web of numerical weights, these models preserve the structure that allows one to trace back to a specific input or condition that led to the result. In theory, this would allow investigators to follow a clearer path from data to decision, something today’s AI engines cannot offer. 

If this promise holds up, it will matter. A military that can show step-by-step why a system chose a target is in a far stronger position to defend that decision legally and ethically. It also strengthens public trust, both domestically and internationally, at a time when the use of autonomous and AI assisted weapons is under growing scrutiny from human rights organizations and international bodies. In this framing, the race for military AI dominance is not only about who can build the fastest or most powerful system. It becomes a race to build systems that are defensible. These systems must withstand the scrutiny of a courtroom, a parliamentary inquiry, or an international tribunal. 

A Promising Idea, Not a Proven One 

It would be a mistake to overstate that this capability is currently accessible. The idea that quantum AI offers built-in decision recovery is still unproven, not an established capability. Techniques used to explain classical AI decisions, tools such as LIME and SHAP, are only beginning to be adapted for quantum models. Early results are encouraging in narrow research settings, but no military or independent certification body has yet accepted quantum-generated decision recovery as sufficient evidence in a legal or operational context. The honest way to describe this field today is as a promising hypothesis worth serious investment, not a solved problem ready for deployment. Overselling it risks the same trap that has plagued classical military AI, where systems are fielded faster than the frameworks needed to hold them accountable. 

Conclusions 

The common denominator in military technology has been about speed. It is about who can evaluate fastest, be more decisive, and often strike first. Computers systems help drive these and AI plays a key role. Quantum AI’s most valuable military application may ultimately have little to do with either. If it can genuinely offer clearer, more transparent reasoning behind life-and-death decisions, it would address one of the most serious gaps in how militaries currently use it. This deserves considerable research attention and scrutiny in equal measures.  

The technology that ultimately proves most valuable on future battlefields may not be the one that thinks the fastest. It may be the one that can provide an answer when someone finally asks why.  

Maheen Butt is a Lahore-based researcher and strategic communications professional conducting research primarily on foreign affairs. She is a Mphil graduate of International Relations from Kinnaird College for Women, and her work focuses on global affairs, emerging technologies mainly on increasing use of AI in military affairs, and defense studies. The Author’s opinions are their own.

Explainability, Not Speed, Will Define the Next Military AI Race was originally published on Global Security Review.

Technological advances and rising military expenditures in recent years have accelerated the development of Lethal Autonomous Weapon Systems (LAWS). Though this technology is still in its infancy, it has already transformed modern warfare. LAWS, when fully evolved, will provide means for precise and independent selection and engagement of targets without exposing soldiers to battlefield dangers. A 2025 Congressional Research Service report titled Defense Primer: U.S. Policy on LAWS classifies it as “a special class of weapon systems that use sensor suites and computer algorithms to independently identify, target and employ an onboard weapon system to engage and destroy it without manual human control.” The US Department of Defense Directive 3000.09, Autonomy in Weapon Systems (2023), defined LAWS as systems that, once activated, “can select and engage targets without further intervention by a human operator.” This concept, known as “human out of the loop” or “full autonomy,” involves target selection and engagement based on inputs from artificial intelligence (AI), big data analytics, and sensor-based identification. 

According to Data M Intelligence, the global autonomous weapons market reached USD 14.2 billion in 2024 and is expected to grow to USD 33.47 billion by 2032, with a compound annual growth rate of 11.39 percent during 2025-2032. Simultaneously, global civil society initiatives are advocating a ban on fully autonomous systems. In October 2012, Amnesty International launched the Stop Killer Robots campaign, an alliance of over 180 organizations across 65 countries, calling for an international law on autonomy in weapon systems to ensure machines are not allowed to make decisions that affect life and death. 

Concerns have arisen over unsupervised use and the potential for system errors that can cause unintended civilian casualties, escalate conflicts, and threaten global peace and security. The increasing integration of autonomous weapon systems in combat has already been highlighted by their reported use in Ukraine conflict and in Gaza. A February 2025 report by the Foundation for Political, Economic and Social Research titled Deadly Algorithms: Destructive Role of Artificial Intelligence in Gaza War revealed that Israel employed AI-based systems, Lavender and Habsora, to identify and attack human targets. The report states that Lavender can approve targets within 20 seconds, often without substantive human review. Since October 2023, the system has compiled a list of 37,000 potential individuals labelled as Hamas members without verifying their military profile.  

Since 2014, the United Nations Convention on Certain Conventional Weapons (UN CCW) has debated the regulation of LAWS. In May 2024, Arms Campaign Director Steve Goose of Human Rights Watch warned that “the world is approaching a tipping point for acting on concerns over autonomous weapons systems,” underscoring the urgency of an international legal instrument. On 2 December 2024, the UN General Assembly adopted Resolution A/RES/79/62 on LAWS by 166 votes in favor, 3 against, and 15 abstentions. The resolution marked a decisive step in acknowledging global concerns over autonomous weapon systems, affirmed the applicability of international humanitarian law (IHL) and called for further consultations in 2025. The first UNGA meeting on autonomous weapons, held on 12-13 May 2025 and attended by 96 countries, including representatives from the International Committee of the Red Cross (ICRC) and civil society, reinforced momentum to prohibit and regulate LAWS. On that occasion, UN Secretary-General António Guterres advocated for a legally binding instrument to ban LAWS by 2026, describing them as “politically unacceptable and morally repugnant.”  

Despite global concerns, progress on a legally binding treaty on LAWS remains elusive due to divergent strategic interests of major powers. The US continues to resist codification of a new binding framework, emphasizing the adequacy of national weapons review mechanisms to preserve strategic and technological flexibility. While the US maintains that it does not currently possess LAWS, senior military leaders have acknowledged that Washington may be compelled to develop them if adversaries do so. Russia has opposed any binding treaty, while China supports negotiations on the CCW and the development of norms “when conditions are ripe.” The European Union, in contrast, advocates for a legally binding international instrument, emphasizing Meaningful Human Control (MHC) and compliance with IHL. The EU’s approach seeks to differentiate between systems that incorporate human oversight and those that operate without it. 

The integration of artificial intelligence into weapon systems also presents an increasing challenge to nuclear deterrence and strategic stability. For instance, during the sidelines of the Asia-Pacific Economic Cooperation (APEC) Summit in Peru in November 2024, the then US President Joe Biden and China’s President Xi Jinping jointly pledged not to integrate AI in nuclear command-and-control systems, recognizing the catastrophic risks of automation in nuclear decision-making. However, as AI rapidly improves surveillance, missile guidance and targeting systems, it is unclear whether this restraint will hold.  

The integration of AI in nuclear forces may introduce instability into deterrence dynamics by reducing decision-making time and increasing challenges caused by algorithmic bias in early warning systems, posing the threat of false nuclear alarms. Cold War history reminds us of human judgment, central to nuclear stability, and averted catastrophes. During the Cuban Missile Crisis, the B-59 submarine incident on 27 October 1962 brought the two superpowers close to nuclear exchange when a Soviet submarine commander considered launching a nuclear-tipped torpedo under the mistaken belief that hostilities had commenced. The refusal by Vasily Arkhipov to authorize the attack prevented a potential nuclear war. Similarly, Stanislav Yevgrafovich Petrov, a lieutenant colonel in the Soviet Air Defense Forces, chose to disregard a false early-warning alert indicating an incoming US nuclear strike in 1983, preventing a global nuclear disaster. Such decision-making underscores the indispensable role of human rationality in nuclear command-and-control systems.  

As LAWS presents multifaceted threats to international peace and security, states need to consider negotiating a legally binding instrument that ensures MHC over autonomy in weapon systems. Enhancing transparency, accountability, and rigorous weapons reviews are essential to prevent destabilization and ensure that technological progress does not outpace the human element in the use of force. Confidence-building measures, such as transparency in military AI, the establishment of international verification mechanisms and a moratorium on the development and deployment of LAWS, could help mitigate future dangers. 

Jawad Ali Shah is a Research Officer at the Center for International Strategic Studies Sindh (CISSS), Pakistan. He holds a BS in International Relations from the University of Sindh, Jamshoro, Pakistan. His research areas are emerging military technologies, and South Asian nuclear deterrence and strategic stability dynamics. The views are the author’s own. 

Lethal Autonomous Weapon Systems: A New Battlefield Reality was originally published on Global Security Review.

That gap undermines American extended deterrence because Washington cannot reassure allies or deter aggressors without accounting for South Asia’s threat calculus. AI arms developments in this region stem from colonial legacies and mistrust of great power intentions, creating a volatile strategic environment.

India’s Governance Innovation in Defense AI

India’s governance model integrates civilian oversight with defense research and ensures ethical deployment of AI. The Responsible AI Certification Pilot evaluated algorithms for explainability before clearance. Its National Strategy for AI mandates ethical review boards for dual-use systems. Developers must document bias-mitigation measures and escalation pathways. Embedding accountability at design phase stabilizes deterrence signals by reducing inadvertent algorithmic behaviors.

The Evaluating Trustworthy AI (ETAI) Framework advances defense AI governance. It enforces five principles: reliability, security, transparency, fairness, privacy, and sets rigorous criteria for system assessment. Chief of Defense, Staff General Anil Chauhan, stressed resilience against adversarial attacks, highlighting the challenge of balancing effectiveness and safety. By mandating continuous validation against evolving threat scenarios, ETAI prevents mission creep and maintains operational integrity under stress.

India’s dual use by design philosophy embeds safeguards within prototypes from inception. This contrasts with reactive models that regulate AI after deployment. Civilian launch-authorization channels separate political intent from technical execution, ensuring decisions remain under human control and reinforcing credibility in crisis moments. Regular red-team exercises involving independent experts further validate system robustness and reduce risks of false positives in autonomous targeting.

Strengthening Extended Deterrence through Cooperation

US-India collaboration on AI verification can reinforce extended deterrence by aligning technical standards and testing protocols. The iCET fact sheet outlines secure information sharing and joint safety trials. Launched in January 2023, iCET has already enabled co-production of jet engines and transfer of advanced drone technologies. Building on this foundation, specialized working groups could develop common benchmarks for adversarial-resistance testing and automated anomaly detection.

A Center for Strategic and International Studies report recommends a trilateral verification cell blending American evaluation tools with India’s ethical reviews. Joint trials of autonomous air-defense algorithms would demonstrate interoperability and resolve. A shared “AI Red Flag” system would alert capitals to anomalous behaviors and reduce strategic surprise. Embedding cryptographically secure logging of decision path data ensures an immutable audit trail for post-event analysis and confidence building.

The INDUS-X initiative, launched during Prime Minister Narendra Modi’s 2023 US visit, integrates responsible AI principles into defense innovation. By aligning standards, both countries ensure AI systems enhance strategic stability rather than undermine it. Expanding INDUS-X to include scenario-based wargaming with allied partners can stress-test ethical frameworks and calibrate thresholds for human intervention under duress. This model can extend under the Quad framework, pressuring authoritarian regimes to adopt transparency measures.

Institutionalizing Global AI Arms Control

A formal arms control dialogue should adopt India’s baseline standards for ethical AI governance. The UNIDIR report calls for universal bias audits and incident-reporting obligations to prevent unintended escalation. Carnegie scholars propose a tiered certification process under a new protocol for autonomous systems within the Convention on Certain Conventional Weapons, requiring peer review of algorithms before deployment. Embedding such certification in national export-control regimes would create global incentives for adherence.

The UN General Assembly has established an Independent AI Scientific Panel and a Global Dialogue on AI Governance to issue annual assessments on risks and norms. This mechanism can evaluate military AI applications and recommend confidence-building measures. Procedural transparency would coexist with confidentiality requirements, balancing security with mutual reassurance. Regular joint workshops on risk-assessment methodologies can diffuse best practices and diffuse mistrust among major powers.

Regional Applications and Future Prospects

India’s responsible AI framework must inspire regional adoption and confidence-building measures. Pakistan and China should engage transparency initiatives to prevent dangerous asymmetries in AI capabilities. Proposed measures include joint research on AI safety, shared performance databases, and collaborative development of detection algorithms.

Successful tests of India’s hypersonic ET-LDHCM system, capable of Mach 8 and a 1,500-kilometer range, underscore the urgency of governance frameworks before fully autonomous weapons deploy. The Quad’s model of Indo-Pacific cooperation provides a template for multilateral norms on responsible AI in defense. Extending these norms to confidence-building measures such as pre-deployment notifications and automated backchannels can reduce the risk of inadvertent escalation.

Looking ahead to the United Nations General Assembly meeting on AI governance in September 2024, American policymakers can leverage India’s experience. Joint verification exercises and an ethical audit regime will establish global norms for military AI. Integrating lessons from ETAI and iCET into the assembly’s resolutions can produce enforceable standards that bind both democratic and authoritarian states. This approach will reaffirm American extended deterrence and help prevent destabilizing AI-driven arms races worldwide.

By demonstrating that ethical AI development strengthens rather than weakens deterrence credibility, India’s model provides both technical solutions and normative frameworks for managing the military applications of artificial intelligence. Sustained international cooperation on these principles is pivotal for securing strategic stability in a rapidly evolving technological landscape.

Vaibhav Chhimpa is a researcher who previously worked with the Department of Science & Technology (DST), India. Views expressed are the Author’s own.

The Artificial Intelligence (AI) Arms Race in South Asia was originally published on Global Security Review.

This was not an isolated incident. Civilian-grade drones and other dual-use technologies are increasingly being used to survey or target public infrastructure. From energy grids to airports, the connective tissue of modern life is exposed to risks once confined to traditional warzones. These developments are reshaping global security policies and blurring the boundary between civilian and military domains.

Civilian Tech, Strategic Impact

Cheap unmanned aircraft systems (UAS) are now accessible worldwide. While drones were initially developed for military use, the most commonly deployed platforms today, such as DJI’s Mavic series, were originally built for civilian applications like aerial photography and videography. Their affordability, portability, and high-spec cameras made them commercially popular, but those same features have made them easy to repurpose for military contexts.

In particular, first-person view (FPV) drones, designed for immersive recreational flying, were rapidly adapted for frontline use in conflict. These drones are now routinely deployed with improvised explosives or used for precision reconnaissance. In Ukraine, both sides repurposed off-the-shelf drones in vast numbers; nearly two million were produced in 2024 alone. Many of these are equipped with AI-enabled navigation and targeting, underscoring how quickly civilian tech can be weaponised.

Non-state actors are following suit. Armed groups are using FPV drones for low-cost, high impact strikes on infrastructure, blurring the lines between military and civilian threats. This second drone age shows that national security vulnerabilities now stem as much from consumer technology as from conventional arsenals.

The broader implication is clear: private-sector innovations, often created without any defense intent, are shaping the battlefield. These companies bring novel use cases, technical advantages, or agile design processes that legacy defense contractors may overlook. Civilian tech is not just a risk; it is a potential strategic asset. Tapping into this ecosystem, especially among start-ups and experts, could redefine how the country protects critical infrastructure in an era of hybrid conflict.

Infrastructure in the Crosshairs

Modern infrastructure is a key target in modern conflicts or hybrid attacks, just like military bases traditionally were. In 2022, after the sabotage of the Nord Stream pipelines, over 70 drone sightings were reported near Norwegian offshore oil platforms. Oslo feared Russian-linked hybrid operations targeting Europe’s energy supply and deployed naval assets and invited NATO allies to assist in patrols.

Meanwhile, Ukraine’s energy grid suffered repeated drone and missile attacks, with waves of low-cost Shahed drones used to disable power plants. By spring 2024, roughly half of Ukraine’s electricity capacity was destroyed, forcing nationwide blackouts.

Outside conflict zones, attacks on infrastructure are also rising. In Sudan, a drone strike on a power station caused regional outages, and other drone attacks on water purification stations left the country on the brink of a significant Cholera outbreak. In the US, federal officials stopped an attack on a power grid by a man using an explosive-carrying drone.

Transportation hubs are vulnerable, too. In January 2025, drone activity shut down Riga Airport, disrupting dozens of flights.

 Gaps in Governance

 Despite growing risks, legal and operational frameworks remain fragmented. Drones and AI-driven surveillance systems often fall outside traditional arms control regimes. As a recent executive order put it, “Criminals, terrorists, and hostile foreign actors have intensified their weaponization of drone technologies, creating new and serious threats to our homeland.”

Jurisdictional confusion is common. In many countries, local authorities lack legal authority to respond to rogue drones above critical sites. Aviation safety rules and privacy laws create hesitation, giving bad actors a head start.

Even when threat awareness exists, coordination is inconsistent. The Cybersecurity and Infrastructure Security Agency warns that drones are used for surveillance and sabotage, yet they lack the comprehensive tools to oversee private-sector resilience or cross-border response.

 A Global Security Challenge

 Drone and AI threats are not confined by borders. In 2023, the European Commission launched a new counter-drone strategy, urging member states to harden infrastructure and coordinate airspace protections. NATO has added counter-UAS exercises to its joint drills, while AUKUS partners are beginning to share emerging drone and AI tactics.

But international law is lagging. There is still no global treaty governing the use of armed drones or autonomous surveillance. Export control regimes struggle to manage proliferation of AI-enabling components. At the UN, efforts to establish binding norms on autonomous weapons are stalled. Ad hoc coordination is, however, slowly improving.

When Norway’s oil platforms were threatened, NATO allies were called in within days. After drone sightings near Dutch and Belgian ports, neighboring governments exchanged countermeasure plans. These models suggest a path forward: rapid and collective responses based on shared tools, shared doctrine, and shared threat intelligence.

The future of civilian dual-use technologies will not be defined by innovators alone. Whether drones or AI software, these tools are already reshaping how adversaries threaten public safety and economic continuity. What is at stake is not just national security, but the resilience of infrastructure that supports daily life.

The Crucial Role of Start-ups in National Defense

 Civilian-origin technologies are now driving the next wave of defense capability. From FPV drones to AI surveillance tools, some of the most disruptive military applications today are emerging not from traditional defense primes but from commercial markets, often developed by start-ups with no military background.

A coordinated international framework is urgently needed, one that does not just support innovation and infrastructure protection but actively integrates civilian tech into defense planning. This means lowering the barriers for experts and start-ups to meaningfully contribute alongside legacy contractors. The United Kingdom’s recent Defence Review hinted at this shift, recognising that smaller firms are vital to national resilience, particularly when civilian infrastructure is under threat.

What is truly needed is a NATO-wide or broader allied framework that enables cross-border collaboration, streamlines regulation, and opens up procurement pathways.

Today, many start-ups working at the intersection of security and technology face steep hurdles: limited access to capital, opaque compliance regimes, and procurement processes designed around, and for, large incumbents. Yet by creating space for their innovation, we can modernize collective defense from the ground up, using the very same civilian tools that adversaries are already turning into weapons.

A coordinated international framework is urgently needed, one that not only supports innovation and infrastructure protection but also lowers barriers to experts and start-ups to contribute more meaningfully alongside traditional defense primes. The UK’s recent Defence Review hinted at this shift, recognizing the value smaller firms bring to national resilience. It is time to take similar action at home.

Harry Geisler is the CEO of YAVA.

How Civilian Dual-Use Technologies Are Reshaping Global Security Policies was originally published on Global Security Review.