Artificial Intelligence and Global Security

Artificial Intelligence and Global Security arrived in 2020 from Emerald Publishing at a moment when defence ministries across the West had begun to treat artificial intelligence not as an emerging capability but as an operational one. Yvonne R. Masakowski, who edited the volume, is a research psychologist associated with the U.S. Naval War College, and much of her career has been spent on how human operators work alongside complex systems under pressure. That grounding shapes the book. Technical descriptions are present, but the questions are framed from the vantage of a defence community that has to make decisions about deployment, doctrine, and training now, rather than a research community writing about capabilities that might exist one day. The essays gather contributors from service research centres, NATO institutions, universities in the United States and Europe, and the wider defence analytical community, and the collection reads as a survey of the questions being asked around policy planning tables rather than laboratory benches.

The central argument that runs through the book is that AI is already a variable in the security environment, not a future one, and that the states which move fastest at integrating it into intelligence, command, and force employment will set the pace for everyone else. Masakowski and her contributors treat AI less as a discrete weapon system and more as a layer sitting beneath almost every domain of modern conflict, from satellite imagery analysis and logistics optimisation to targeting, cyber operations, and the tempo of decision-making itself. The strategic implication follows quickly: adoption speed and organisational adaptation, rather than the underlying algorithms, will separate the powers that gain an edge from those that fall behind. This is a claim aimed at Western defence planners who might still be treating machine learning as a research programme rather than a procurement priority, and it is stated with a directness that reflects the operational anxiety of its authors.

The book joined a rapidly growing shelf of policy-oriented AI-in-security volumes that appeared in the late 2010s and early 2020s, at a time when the U.S. Department of Defense had stood up the Joint Artificial Intelligence Center, when China had published its New Generation Artificial Intelligence Development Plan and begun weaving intelligentized warfare into official doctrine, and when Russia had made highly visible investments in unmanned ground and aerial systems while its president publicly declared that whoever led in AI would lead the world. The volume did not attempt to be a technical primer. Readers who wanted an introduction to neural network architectures, transformer models, or the mathematics of reinforcement learning would need to look elsewhere. What Masakowski assembled instead was a strategic-operational survey, and the book fits alongside the volumes coming out of RAND, the Center for a New American Security, the Stockholm International Peace Research Institute, and the various NATO Centres of Excellence during the same window.

The structure moves outward from the technology to its consequences. Early chapters set out what AI is doing in the security domain now, with attention to machine learning for intelligence exploitation, autonomy in unmanned systems, and the ways algorithmic support is beginning to compress the observe-orient-decide-act loop. From there the book moves through weapons and platforms, into cyber and information operations, into the great-power competition frame, and then into questions of governance, ethics, and the human operator. This progression mirrors the way defence organisations themselves have tended to encounter the field: a capability appears, it produces an operational question, the operational question surfaces a policy question, and the policy question runs into a legal or ethical one.

The concrete material is where the book earns its keep, and several themes recur across chapters. Autonomous and semi-autonomous weapons receive sustained attention, with contributors returning to the debate that has run for years at the United Nations Convention on Certain Conventional Weapons in Geneva, where a group of governmental experts has met to consider lethal autonomous weapons systems without producing binding rules. The book treats this debate with a working assumption that a total ban is unlikely and that the meaningful questions concern meaningful human control, testing and evaluation, and the doctrine surrounding use rather than prohibition. Readers get walked through the distinction between autonomy in navigation and autonomy in engagement, and through the way loitering munitions, unmanned aerial systems, and torpedo and mine warfare are already blurring the line long before any formal debate concludes.

Machine learning applied to intelligence is another recurrent theme, and the book takes the view that the volume and heterogeneity of modern collection — commercial satellite imagery, signals, social media, financial data — long ago exceeded what human analysts can process. Automated triage, anomaly detection, pattern-of-life analysis and image recognition are described as capabilities that intelligence services either adopt or fall behind on. Command and control receives a parallel treatment, with contributors describing decision-support systems that surface options faster than a staff can, and warning that the temptation to trust an opaque recommendation grows with the tempo of the fight. Deterrence theory in an age of machine speed features prominently: several chapters examine what happens to crisis stability when reaction times shrink from hours to minutes and when adversaries cannot be sure whether an action was authorised by a person or by a model. The classical deterrence literature was written for humans considering escalation with full political weight; the book asks what happens when the human is one step further back.

Great-power competition is treated as the strategic frame in which everything else sits. Chapters compare how the United States, China, and Russia are pursuing military AI, and the comparison is not neutral. China is described as pursuing intelligentized warfare with a whole-of-nation approach that fuses commercial AI capacity, dual-use civilian research, and a stated ambition to leapfrog rather than match. Russia is described as more constrained by its industrial base and yet aggressive about unmanned platforms, electronic warfare, and information operations, and the book was written before the full-scale invasion of Ukraine but with awareness of what Russian doctrine and capability development had been signalling for years. Iran and North Korea appear in the margins as states whose interest in cheap autonomous systems and cyber capabilities makes them relevant despite their modest budgets. NATO comes in for its own attention, with contributions considering how an alliance of very different national capabilities is meant to coordinate around a technology that a small number of members lead in and the rest depend on.

Information operations and cognitive security form another current in the book, and this is one of the more prescient sections in hindsight. Deepfakes, synthetic media, and AI-enabled disinformation are discussed as instruments in a broader competition over the perceived legitimacy of political systems, and the book anticipates the direction in which this problem has since travelled. It is careful to keep this a discussion of statecraft rather than a discussion of any specific election or political actor, and the observational tone reads as an attempt to describe a manoeuvre space rather than to score any partisan point. Cyber, treated in parallel, is discussed as both a domain where AI accelerates defence and one where it accelerates offence, with detection speed and adversary tooling engaged in a familiar cat-and-mouse pattern.

The thread that most obviously reflects Masakowski’s own scholarly background is human-machine teaming. Several chapters examine trust calibration between operators and systems, the cognitive load imposed by working alongside autonomous partners, the risk of automation bias, and the training pipeline needed to produce a warfighter who neither refuses to use a system nor defers to it uncritically. This section reads as a reminder that adoption is not simply a matter of buying software. The organisational, educational, and psychological demands on the humans in the loop are described as the constraining factors, and the book suggests that the side which handles them well will convert its technology into effectiveness faster than the side that focuses only on procurement.

The ethical and legal chapters return to questions that have not been settled by the field and probably will not be settled by any single volume. What does meaningful human control mean in practice for a system that must react in milliseconds? Where does responsibility sit when an autonomous system produces an outcome that a human commander did not specifically authorise? How should the laws of armed conflict be applied to targeting recommendations generated by an opaque model? The book does not pretend to have final answers. It offers the arguments in play, tries to make them intelligible to a defence audience without dumbing them down for a legal one, and lets the reader see where the honest disagreements sit.

Reception in the field has been what one would expect for a multi-author edited volume from a specialist publisher: it entered the reading lists of war colleges and security studies programmes, it has been cited in policy work coming out of think tanks writing on autonomy and human factors, and it has been used as a starting point rather than a destination. The counterpoints raised by other authors since publication have tended to converge on two themes. The first is that any survey of the AI-in-security landscape ages rapidly, because the underlying capabilities do. Chapters written before the mainstream arrival of large language models could not anticipate the specific ways generative AI would reshape open-source intelligence, information operations, or command decision aids. The second is that a book edited from within a U.S. defence education institution inevitably tilts towards a U.S. and NATO-centric account of the field, and that other perspectives — Ukrainian battlefield experience, Israeli operational lessons, Turkish and South Korean defence-industrial trajectories — either arrived after publication or were understandably not the focus. Both criticisms are fair, and both are the kind of criticism that any similar volume would attract.

Placed on the shelf next to its neighbours, the book works as a structured tour of the questions defence ministries and general staffs are being forced to answer. It sits well alongside Paul Scharre’s work on autonomy in weapons, alongside the Congressional Research Service and RAND primers on military AI, and alongside newer treatments of intelligentized warfare and drone-heavy conflict. For a mid-career military officer working towards a war college seminar paper, for a graduate student in security studies looking for a way into the debate, and for a defence analyst trying to keep up with a field that will not stop moving, the volume delivers a coherent map. For a general reader with an interest in geopolitics and defence, the multi-author format and the assumed familiarity with strategic vocabulary make it denser than a single-author treatment would be. The individual chapters are short enough to be read in isolation, which suits a reader who wants to sample rather than commit.

The parts of the book most likely to age well are the human factors sections, the framing of AI as a layer rather than a weapon, and the argument that adoption speed is the decisive variable. Those observations are not tied to a specific model or a specific system, and they remain load-bearing five years on. The parts most obviously dated are the capability snapshots. The pace of change in generative AI, in commercial imagery, in inexpensive strike and reconnaissance drones, and in battlefield electronic warfare has been faster than any book published in 2020 could have captured. The subsequent war in Ukraine has demonstrated, in the plainest possible way, how rapidly a peer competitor to Western defence establishments can develop, field, and iterate autonomous and semi-autonomous systems under fire. A reader coming to Masakowski’s volume today should treat it as the map of a landscape whose contours it correctly identified and whose specific features have kept moving, which is the honest situation for anyone writing seriously about AI in war.