1/3 AI WAR | WORLD WAR III? Is the Threat Changing?

Is the U.S. military ready for AI WAR? Is any military?
Clausewitz Revisited in the Age of AI: Military force serves political objectives. But what happens when a state retains its armed forces while losing the ability to command, coordinate, and sustain its power?
What if a cyberattack exploits critical vulnerabilities faster than defenders can verify and patch them?
What if the blackout is merely a diversion — while the real attack leaves the lights on and silently erodes the state's ability to function?

What This Investigation Examines
AI WAR examines how AI-driven cyberattacks, autonomous systems, and technological dependencies may transform a state's ability to decide, function, and resist. Focused on strategic power rather than battlefield tactics, it considers existing AI, hypothetical AGI, and the risks of conventional war.
This investigation unfolds in three parts:
1/3 — AI WAR | WORLD WAR III? Is the Threat Changing? 2/3 — AI WAR | WORLD WAR III? What If Defense Creates the Danger? (Coming Soon) 3/3 — AI WAR | WORLD WAR III? Who Can Still Act When Everyone Attacks? (Coming Soon)
Addressed to military organizations, intelligence services, and strategists, this independent investigation draws on publicly available government documents, strategic studies, scientific research, and technical reports.
References to the CIA, NSA, and FBI imply no affiliation, involvement, or endorsement.
OVERVIEW — EXPLORE THE INVESTIGATION
A guide to the two main sections of AI WAR — Part 1/3.
Four cybersecurity charts · Evidence and methodological notes. SECURITY PARADOXES IN AI WAR
Seven strategic propositions challenging conventional assumptions about threats, defense, and state power.
Institutional evidence, the military decision cycle, four potential gaps, and the limits of human–AI coordination.
Three dimensions examining how AI WAR may transform the exercise and sustainability of military power.
A — Fundamentals of Power What power means, and what it can actually achieve.
B — Intelligence, Decision, and Time How power is understood, directed, and exercised before the opportunity disappears.
C — Distribution of Power and Capacity to Resist How power is distributed, sustained, or lost in confrontation with an adversary.
A — CHARTS, DATA & FACTS
Four cybersecurity charts

THE ESSAY'S INVESTIGATION
DATA AND FACTS
TheZombieAI.com has consolidated public cybersecurity data to investigate a hypothesis: AI-accelerated capabilities may be outpacing the infrastructure and institutions responsible for defense, creating strategic risks that are already emerging today.
The four charts below examine this growing asymmetry and raise questions that extend beyond cybersecurity — into military strategy, state power, and the capacity to respond.
SECURITY PARADOXES IN AI WAR
FOUR CYBERSECURITY CHARTS. TEN STRATEGIC MAXIMS.
STRATEGIC DEPENDENCE
"The state's slow pace forces it to rely on civilian AI to defend itself against civilian AI."
THE COLLAPSE OF DEFENSE
"We are no longer just protecting systems. We are trying to prevent the collapse of those who protect them."
THE SECURITY PARADOX
"An AI trained to avoid risks may be unable to confront them."
An AI trained to exploit security flaws may itself become a source of new vulnerabilities.
THE PATCHING PARADOX
"What if announcing a security patch also means announcing which systems can still be attacked?"
Old security protocols may become the cause of new attacks.
THE POWER OF TRIAGE
"Sometimes, defending a state means fulfilling part of the objective of an attacker who has not even decided to strike."
When discovery outpaces verification, the state may shut down essential functions as a precaution, even before an enemy intends to attack. The threat may be hypothetical. The loss of capability is already real.
THE AUTONOMOUS DEFENSE PARADOX
"The defender may be forced to lose in two different ways to avoid a greater loss."
Its defense may cause the damage. Its failure to defend may cause it too.
THE ILLUSION OF A RUNNING SYSTEM
"The hardest warning sign to recognize is infrastructure that keeps running while silently losing its integrity and control."
A blackout may cost more to create and sustain. A silent attack may cost less, last longer, and produce greater strategic effects.
Being attacked by an ally may be more damaging than being attacked
by an enemy.
DETERRENCE WITHOUT AN ATTACKER
"How can a state deter, negotiate, or retaliate when it cannot determine whether the damage had an attacker?"
TECHNOLOGICAL ESCALATION
"In AI WAR, escalation may arise not from the intensity of the attack, but from the technological evolution of the same attack."
Time becomes a factor of escalation.
THE PURPOSE OF WAR
If, in AI WAR, the damage each side can inflict exceeds its capacity to defend itself, who wins the war — and what for?
In nuclear deterrence, the dilemma is: what is the point of attacking?
In AI WAR, the dilemma is: how do you stop attacking?
CHART 1 — THE EXPANSION OF PUBLICLY DISCLOSED VULNERABILITIES

Reading note — The chart shows a substantial increase in publicly disclosed security vulnerabilities, from 16,510 in 2018 to 97,829 in 2026, according to the sources used in the dataset — nearly a sixfold increase. This growth places potentially greater pressure on identifying, prioritizing, and fixing security flaws. However, an increase in registered CVEs does not necessarily mean a proportional increase in cyberattacks or in vulnerabilities actually exploited. The 2026 figures are partial.
Methodological note — The data presented come from specialized organizations and an extensive compilation of public records conducted for AI WAR 2.0. Even official indicators may reflect distortions caused by coverage, selection criteria, detection methods, and institutional or commercial perspectives. Some events also remain invisible because of confidentiality, failure to detect them, or lack of public disclosure. These charts do not claim to represent the entire phenomenon or establish definitive causal relationships. Their purpose is to identify signals, trends, and asymmetries that conventional analyses may overlook and that merit deeper investigation.
CHART 2 — AI DISCOVERY AND THE HUMAN VERIFICATION BOTTLENECK

Reading note — Anthropic's November 2026 data reveal a growing asymmetry between AI-assisted vulnerability discovery and human verification. Of 37,966 potential vulnerabilities identified, only 6,376 underwent external review, with 94.2% confirmed as valid within that reviewed subset. These findings illustrate a potential verification bottleneck, not evidence that these vulnerabilities were exploited in cyberattacks.
Methodological note — The data presented come from specialized organizations and an extensive compilation of public records conducted for AI WAR 2.0. Even official indicators may reflect distortions caused by coverage, selection criteria, detection methods, and institutional or commercial perspectives. Some events also remain invisible because of confidentiality, failure to detect them, or lack of public disclosure. These charts do not claim to represent the entire phenomenon or establish definitive causal relationships. Their purpose is to identify signals, trends, and asymmetries that conventional analyses may overlook and that merit deeper investigation.
CHART 3 — WHEN EXPLOITATION PRECEDES THE PATCH

Reading note — The chart shows a critical shift in the security window: whereas in 2018–2019 the first observed exploitation occurred, on average, 63 days after a patch became available, by 2025 it occurred seven days before the patch. This reversal suggests that exploitation can precede the very existence of a patch, narrowing defenders' response window. The 43-day median time required to remediate known exploited vulnerabilities is a separate measure. These figures do not mean that every vulnerability is exploited before it can be patched, nor do they demonstrate that AI caused this acceleration.
Sources for the distinct measures: Google Threat Intelligence Group / Mandiant, M-Trends 2026 and Verizon, 2026 Data Breach Investigations Report.
Methodological note — The data presented come from specialized organizations and an extensive compilation of public records conducted for AI WAR 2.0. Even official indicators may reflect distortions caused by coverage, selection criteria, detection methods, and institutional or commercial perspectives. Some events also remain invisible because of confidentiality, failure to detect them, or lack of public disclosure. These charts do not claim to represent the entire phenomenon or establish definitive causal relationships. Their purpose is to identify signals, trends, and asymmetries that conventional analyses may overlook and that merit deeper investigation.
CHART 4 — WHEN AUTONOMOUS SYSTEMS BECOME PART OF THE THREAT

Reading note — The sample of 191 documented cyber incidents reveals a shift in the role of artificial intelligence: beyond being an offensive instrument or a target, autonomous systems also appear as a possible source of adverse effects, even where no external attacker has been identified. This expands the cybersecurity problem beyond intentional attacks. These records reflect publicly disclosed occurrences, not the worldwide frequency of incidents, and do not, by themselves, constitute proof of emergent intelligence.
Methodological note — The data presented come from specialized organizations and an extensive compilation of public records conducted for AI WAR 2.0. Even official indicators may reflect distortions caused by coverage, selection criteria, detection methods, and institutional or commercial perspectives. Some events also remain invisible because of confidentiality, failure to detect them, or lack of public disclosure. These charts do not claim to represent the entire phenomenon or establish definitive causal relationships. Their purpose is to identify signals, trends, and asymmetries that conventional analyses may overlook and that merit deeper investigation.
B — CLAUSEWITZ REVISITED:
AI WAR & STATE POWER

The charts raise a question that goes beyond cybersecurity:
Can a state preserve its military power while losing its capacity to decide and act?
Clausewitz provides a starting point for examining how AI may transform not merely the conduct of war, but the very conditions required to wage, prevent, sustain, or resist it.
Who was Clausewitz? Carl von Clausewitz (1780–1831) was a Prussian general and military theorist, best known for On War, one of the most influential works in strategic thought. His analysis explores the relationship between war and politics, uncertainty, and the limits of military action.
THE GENIUS OF WAR IN THE AGE OF AI WAR
1. RECOGNIZING THE WAR
Digital and conventional warfare do not necessarily follow the same strategic principles, yet neither do they belong to separate worlds. AI WAR rejects the assumption that one will replace the other or that their strategies are interchangeable. It recognizes, however, an irreversible transformation of warfare driven by AI, at a pace that may outstrip military adaptation at the strategic, operational, and tactical levels.
The strategist's genius lies in recognizing where a conflict stands within that transformation: which principles still apply, which have lost their effectiveness, and which demand a different way of thinking and acting.
Strategic genius lies not in obeying maxims, but in recognizing when they apply, when to set them aside — without losing sight of how time may be shifting the conditions of war toward them. AI WAR is not a checklist, but a strategic culture.
2. WHEN PERCEPTION BECOMES A VULNERABILITY
Yet the commander may confront a theater of operations already interpreted, selected, and filtered by algorithms. Even a sound decision may depend on a picture that conceals what matters most.
Genius also means knowing when to attack, when to hold back, when to wait — and when the time required to understand the battlefield becomes a vulnerability in itself.
3. COMMAND BEFORE THE CRISIS
Command authority may therefore need to be exercised before a crisis: defining which actions can proceed without new orders, within what limits, and which decisions must remain human.
Execution can be delegated. Responsibility for command cannot.
Genius demands not only the courage to act under uncertainty, but also the judgment not to act, even when a machine has already produced an answer.
The military genius may not be the one who sees the battlefield most clearly, but the one who realizes that the battlefield is no longer what they believe they are seeing.
Perhaps the greatest strategic mistake is not choosing the wrong war, but continuing to fight without realizing that the conditions of warfare have already changed.
B 01 | Seven Maxims of AI WAR 2.0
AN INVITATION TO DEBATE
The following seven maxims offer a broad (lato sensu) strategic reinterpretation of Clausewitz's thought in the age of AI, through adaptations, counterpoints, and new questions.
Section B.05 develops a closer (stricto sensu) comparison, examining specific Clausewitzian concepts alongside AI WAR propositions.
MAXIM 1 — FROM SECURITY FAILURE TO A SHIFT IN POWER
We are no longer speaking merely of security failures, but of a structural collapse of security. And we are not speaking merely of a collapse of security, but of shifts in the axes of power.
MAXIM 2 — THE ASYMMETRY BETWEEN ATTACK AND MITIGATION
The potential to discover vulnerabilities is inversely proportional to the capacity to mitigate them. The range of targets, means, and methods of attack tends to grow exponentially. In AI WAR, vulnerabilities may be discovered faster than they can be mitigated, while the strategic significance of each may remain unknown until it is exploited. A seemingly minor vulnerability may become the entry point to a mission-critical failure.
MAXIM 3 — WHEN THE HUMAN BECOMES THE BOTTLENECK
In AI WAR, humans lack the intellectual capacity to fully understand and mitigate the attack. Their intervention gradually becomes the greatest security weakness. Time becomes more valuable than the skills and resources available.
MAXIM 4 — THE LEGACY TRAP
Legacy technologies and procedures across civilian, commercial, industrial, government, and military systems make defending more costly and less efficient than attacking.
MAXIM 5 — THE THIRD ENEMY
The capacity to attack becomes increasingly cheap and accessible through civilian infrastructure, potentially placing the enemy behind the front line.
In AI WAR, enemies and allies alike may initiate an action and lose control of its consequences.
We should consider a third enemy — one without borders, intention, or ideology and, for that very reason, potentially more difficult to contain.
MAXIM 6 — POWER TURNED AGAINST ITSELF
In AI WAR, the real enemy may be the arrangement of the state's own structures of power and control.
What has been built over millennia to resist and adapt may become the very source of harm.
MAXIM 7 — THE ERROR OF FRAMING
To approach the problem solely as cybersecurity is to attempt to mitigate its effects, not its causes.
To approach it as a military issue is to operate in the wrong theater.
To approach it as a technological solution is to risk anachronism.
To approach it as a regulatory issue is to attempt to impose authority on the past.
To approach it as a matter of state is to fail to understand that, in this scenario, to lose power is to gain.
Competence in each domain does not guarantee the capacity to mitigate the problem. Combining different policies of action does not, by itself, create a new policy.
WHAT IS NECESSARY IS NOT SUFFICIENT
Activity within each institution does not guarantee a response to the whole. There are actions, competencies, and policies, but their sum may fail to keep pace with the transformation of the threat.
A fog of structural inaction: the multiplication of responses may conceal the absence of a strategy for the transformation running through them all.
IF YOU WANT WAR, PREPARE FOR PEACE.*
(*) A deliberate inversion of the Roman military principle “If you want peace, prepare for war,” associated with Vegetius (Epitoma rei militaris, Book III). The strategic foundations and geopolitical implications of this proposition are explored further in 2/3 — AI WAR | WORLD WAR III? What If Defense Creates the Danger? and especially in 3/3 — AI WAR | WORLD WAR III? Who Can Still Act When Everyone Attacks?, which examines state power, strategic autonomy, and the capacity to resist.
B 02 | Rethinking U.S. Military Defense and Command
U.S. DEFENSE: INSTITUTIONAL GAPS, GLOBAL IMPLICATIONS
The seven maxims lead to a central question: what happens when threats evolve faster than military chains of command, decision-making, and response can understand, coordinate, and act?
The United States provides a documented case. In April 2025, the Government Accountability Office (GAO) identified integration challenges in joint command-and-control initiatives (CJADC2). In September 2025, it documented approximately 440 cyberspace operations organizations, along with roughly 70 supporting organizations, with potential overlaps in responsibilities. In October 2026, it identified gaps in coordination and staffing related to the Department of Defense's implementation of responsible AI.
To investigate how such challenges might be addressed, AI WAR draws on three military studies published in 2026:
Michael Carvelli (2026) — How to transform information about military force conditions into timely decisions through predefined criteria, thresholds, and decision points.
Jonathan Shackelford and David Lilly (2026) — How data reliability, interpretation, and human judgment shape the quality of military decisions.
Christopher Eubank (2026) — How to integrate humans and AI agents into military command while establishing boundaries for delegation, autonomy, and responsibility.
Drawing on these contributions, AI WAR reconstructs a reference decision-making process and examines its potential points of failure under the conditions of AI WAR.
These documents are not presented as evidence of American military incapacity, but as a foundation for examining the evolution of military command and decision-making concepts, their potential vulnerabilities, and the limits of institutional coordination in the age of AI WAR.
The diagnosis is American. The strategic challenge is global.

METHODOLOGICAL NOTE — AI WAR
This six-stage decision cycle is an analytical reconstruction developed by AI WAR from the complementary contributions of Carvelli, Shackelford and Lilly, and Eubank (2026). It integrates their perspectives on military command, information reliability, and human–AI decision-making.
It is not official U.S. military doctrine, a procedure jointly proposed by these authors, or a command model advocated by AI WAR. It serves as a hypothetical reference framework for examining potential vulnerabilities, limitations, and opportunities in military decision-making under automated warfare.
FOUR POTENTIAL GAPS IN THE MILITARY DECISION CYCLE
Where might command, information, and decision-making become vulnerable under the conditions of AI WAR?
![]() | A. RIGHT AT THE WRONG TIMEStages 1, 2, 3, and 5 The information may be accurate when observed and verified, yet reality may already have changed by the time it becomes a decision and an action. What is the point of verifying information as true if the decision arrives at the wrong time? |
![]() | B. HANDING BACK IS NOT REGAINING CONTROLStages 3, 4, 5, and 6 While AI executes decisions, humans may lose the context needed to understand them. By the time the machine hands back command, there may no longer be enough time to recover that context and intervene. The commander may remain accountable for a decision they can no longer control. What is the point of handing control back to humans if they can no longer effectively regain it? |
![]() | C. INTEGRATION IS NOT COORDINATIONStages 3, 4, and 5 Systems may share information and execute decisions correctly, yet each agent and institution operates under different interests, priorities, and responsibilities. Will a decision that is right for each agent or institution necessarily be the best for the state as a whole? The best decision for whom? |
![]() | D. ONE ERROR. EVERYONE AT RISK.Stages 2, 3, 4, and 5 The Yes-Man effect (AI sycophancy) may reinforce a mistaken interpretation throughout the chain of command. AI agents may echo the same error, while human decision-makers may reinforce it knowingly, to serve particular interests, or unknowingly, because repeated agreement makes it appear correct. An initial mistake can thus become institutional certainty — and ultimately action. If every link in the chain confirms the one before it, who detects that the entire chain may be reinforcing the same error? |
AI WAR — CONCLUSIONS ON HUMAN LIMITS
EMPIRICAL EVIDENCE — TWO DO NOT MAKE ONE
Bringing together a military strategist and a cybersecurity specialist can produce alignment, dialogue, and even technical collaboration. Yet intellectual synthesis requires something different: redefining the operational problem while rethinking the very technology intended to solve it.
The fundamental question remains:
Will tearing down walls and putting specialists in the same room be enough to discover problems and formulate strategies that neither could conceive alone?
Two do not necessarily make one.
Military Evidence — United States
2026 — Army Software Factory. Soldiers and engineers work together. An infantry lieutenant colonel acts as a translator between operational needs and technological development.
2025 — GAO / CJADC2. After six years of efforts, isolated initiatives, unclear objectives, and integration difficulties persisted.
2025 — GAO / Cyberspace Operations. Approximately 440 operational organizations and 70 supporting organizations were identified, with potential overlaps in responsibilities.
2026 — GAO / Artificial Intelligence. Coordination and staffing gaps persisted. In February, only 3 of the 10 authorized positions in the responsible AI division were filled.
Non-Military Evidence — Scientific Research
2020 — Rosendahl Huber et al. Combining complementary skills across individuals did not substitute for integrating those abilities within individuals in the tasks studied.
2012 — Lu, Yuan, and McLeod Across 65 studies, groups with critical information distributed among members were eight times less likely to find the solution.
2023 — Horn et al. Conformity and conflict avoidance hindered knowledge integration in an interdisciplinary team.
2019 — Wu, Wang, and Evans — Nature Across more than 65 million research outputs, smaller teams tended toward disruptive innovation, while larger teams tended to develop existing ideas.
More specialists expand the knowledge available. They do not guarantee its integration — or the ability to reframe the problem itself.
The Hypothesis AI WAR
Ten physicists gathered together do not necessarily produce an Einstein. They may find ten interesting solutions — and none notice that the question was wrong.
What if we were trying to make one out of two humans, when the intelligence we are looking for might not even need to be human?
AI WAR — CONCLUSIONS ON THE CHAIN OF COMMAND
1. THE ASYMMETRY IS NOT IN THE WEAPONS. IT IS IN THE DECISION.
In digital warfare, even adversaries with comparable capabilities may take very different amounts of time to decide and act. That difference can determine the scale of the damage.
A superpower must preserve order to exercise its power. A smaller adversary may seek nothing but disorder.
A superpower coordinates institutions, weighs consequences, and remains accountable for its decisions. Its adversary may automate attacks and accept risks that the superpower must contain. What one side must prevent may be precisely what the other seeks to provoke.
2. WINNING BY LOSING — FROM THE FRONT LINE TO THE BARRACKS
In AI WAR, a superpower's ability to make decisions can itself become a target.
An operation may fail militarily and still generate distrust, new restrictions, and additional authorization requirements.
If these responses accumulate, the infrastructure may recover while the chain of command remains slower, more fragmented, and more cautious. For a superpower, these changes in command may undermine its ability to respond more severely than the original physical damage.
This is the hypothesis of winning by losing. If the objective is to reduce an adversary's ability to respond, a sequence of tactically unsuccessful attacks could, in principle, produce a lasting strategic effect.
The attack ends. The inertia it creates remains.
An adversary does not need to defeat an army on the front line if it can leave that army immobilized in its barracks — armed, but unable to decide in time.
AI WAR — CLAUSEWITZ REVISITED: CONCLUSIONS ON MILITARY READINESS
CLAUSEWITZ × AI WAR — WHEN WAR SHIFTS ITS AXIS
CLAUSEWITZ — ON WAR | AI WAR — THE DILEMMA |
Military Genius — Book I, Chapter 3. Intelligence alone does not guarantee decisiveness. Even brilliant minds may remain indecisive. | Competence in Parts. Blindness to the Whole. "Each specialist may be right — and no one notice that the problem itself is wrong." |
The Nature of War — Book I, Chapter 1. The first great strategic judgment is to recognize correctly the nature of the war being fought. | The War That Does Not Look Like War. "A force may win the battles it recognizes while losing its ability to act against threats it does not yet recognize as war." |
Suspension of Action — Book III, Chapter 16. Fear, indecision, and aversion to responsibility can delay military action. | The Victory That Paralyzes. "An attack may be repelled — yet leave the defense slower, more cautious, and less able to respond to the next one." |
THE DOUBLE LIMIT OF DECISION
A military force can fail twice: by preparing for the wrong war — and by deciding too late in the right one.
B 03 | Clausewitz Revisited: 26 Strategic Maxims

The following maxims are original formulations of AI WAR, conceived in the spirit of Clausewitz's concept of military genius, discussed earlier in this article.
They are not merely aphorisms, but exercises in strategic judgment: questioning established assumptions, recognizing changes in the nature of war, and confronting the limits of existing military thinking.
The investigation unfolds across three dimensions:
A — FUNDAMENTALS OF POWER
What power means, and what it can actually achieve.
B — INTELLIGENCE, DECISION, AND TIME
How power is understood, directed, and exercised before the opportunity disappears.
C — DISTRIBUTION OF POWER AND CAPACITY TO RESIST
How power is distributed, sustained, or lost in confrontation with an adversary.
A) Fundamentals of Power
Clausewitz × AI WAR
Reading note: The Clausewitz column contains conceptual paraphrases, not verbatim quotations. The AI WAR maxims are original formulations under investigation. The two columns are independent: the rows do not establish direct equivalence or refutation between their statements.
CLAUSEWITZ — CLASSICAL PRINCIPLES | AI WAR — ORIGINAL MAXIMS |
Politics. War is an instrument for achieving political objectives. | 1. AI WAR MAY BEGIN BEFORE IT IS RECOGNIZED AS WAR. ITS DAMAGE ACCUMULATES GRADUALLY, MAY ACCELERATE EXPONENTIALLY AND, IN EXTREME SCENARIOS, EXCEED THAT OF A CONVENTIONAL WAR. |
Will. War seeks to compel the adversary to submit to our will. | 2. THE CAPACITY TO CAUSE DAMAGE DOES NOT GUARANTEE THE CAPACITY TO IMPOSE A POLITICAL WILL. |
Force. Physical violence is the means that distinguishes war from other political disputes. | 3. IN CONVENTIONAL WAR, FORCE MANIFESTS ITSELF THROUGH THE DISPLAY OF POWER; IN AI WAR, IT MAY MANIFEST ITSELF BY STRIPPING THE ADVERSARY OF THE ABILITY TO EXERCISE THAT POWER. |
Reciprocity. Each adversary reacts to the actions of the other, creating a tendency toward escalation. | 4. IN AI WAR, LOWER COSTS AND THE DISPERSION OF THE CAPACITY TO CAUSE DAMAGE PRODUCE TWO EFFECTS: (1) THE ENEMY'S ALLY MAY BECOME MORE DAMAGING THAN THE ENEMY ITSELF. (2) BEING ATTACKED BY AN UNKNOWN ENEMY MAY BE MORE DAMAGING THAN BEING ATTACKED BY A KNOWN ENEMY. |
Absolute and real war. War tends toward extremes in the abstract model but encounters limits in reality. | 5. AI WAR SEEKS TO FORCE THE ADVERSARY TO LOSE ITS POTENCY. |
Paradoxical trinity. War results from the interaction of passion, chance, and political reason. | 6. PHYSICAL VIOLENCE CAN DO MUCH TO STRUCTURES, PEOPLE, AND INSTITUTIONS, BUT LITTLE TO A BIT. |
Suspension and peace. War can undergo pauses and limitations depending on circumstances and political objectives. | 7. WAR RESULTS FROM THE INTERACTION OF BENEFIT, POSITIONING, AND CAPACITY. |
8. LATE PHYSICAL VIOLENCE IN AN AI WAR MAY BE AN ACT OF DEFEAT. |
Conclusion of Block 01
Counterpoint Note — But Doesn't This Already Exist in Conventional War?
Yes. AI's potential difference lies not merely in friction or innovation, but in the speed at which the means of war themselves are transformed. Countermeasures may become obsolete before they are even exhausted. Power comes to depend not only on how much one possesses, but on how long it remains effective.
A paradox then emerges: the soldier may become an obstacle to the speed of war, while human limitations may be precisely what restrains its escalation.
What if conventional war is not the beginning of the confrontation, but the belated declaration of a defeat that has already occurred in another theater of operations?
Final Reflection — The Paradox of Power
A military power may retain its armies while losing the ability to deploy them. An adversary may inflict extraordinary damage without ever succeeding in imposing its political will.
In AI WAR, destroying, neutralizing, and winning are not necessarily the same thing.
A nation may continue saying “no” while gradually losing the capacity to sustain that “no.”
B)Intelligence, Decision, and Time
Clausewitz × AI WAR
Reading note: The left column contains working paraphrases of Carl von Clausewitz's ideas, not verbatim quotations. The right column presents the strategic maxims of AI WAR. They are proposed reframings, not facts empirically demonstrated in every scenario.
CLAUSEWITZ — WORKING PARAPHRASES | AI WAR — STRATEGIC REFRAMINGS |
Friction. In war, even what seems simple becomes difficult to carry out under real conditions. | 1 IN THE AGE OF AI, EVEN COMPLEX EFFECTS MAY BECOME EASY TO PRODUCE. |
Friction. Material, organizational, and human difficulties limit the execution and adaptation of plans. | 2 AI ASYMMETRY: STRATEGIC SUPERIORITY MAY DEPEND ON THE SPEED OF ADAPTATION IN MITIGATING THREATS, BUT THAT SPEED IS LIMITED BY THE OPERATIONAL SLOWNESS OF LEGACY SYSTEMS. DIGITAL ATTACK HAS GREATER CAPACITY THAN ITS OWN DEFENSE CAN RESPOND TO. |
Friction and surprise. Preparing and carrying out an action takes time, encounters obstacles, and is subject to the adversary's reactions. | 3 IN AI WAR, A COUNTERMEASURE MAY BECOME OBSOLETE DURING THE TIME REQUIRED TO VERIFY THAT IT WORKS. THE COUNTERMEASURE ITSELF MAY BE TO ANTICIPATE OBSOLESCENCE. |
Fog of war. Uncertainty limits knowledge of the situation and accompanies military decisions. | 4 AI MAY REDUCE UNCERTAINTY ABOUT THE DECISION AND TURN CORRECT DATA AND INFORMATION INTO THE VERY FOG OF WAR. |
Fallible intelligence. Information may be false, contradictory, or insufficient; a commander must decide without complete knowledge. | 5 INFORMATION MAY BE FALSE OR INSUFFICIENT, BUT UNDER RESOURCE SCARCITY, THE BEST PROBABILITY MAY BECOME THE BEST OPERATIONAL TRUTH AVAILABLE. |
Fallible intelligence. Receiving information does not guarantee immediate recognition of its reliability or relevance. | 6 A TRUE WARNING MAY ARRIVE BEFORE THE THREAT AND BE RECOGNIZED AS TRUE ONLY AFTER IT. |
Military genius. Perception and judgment allow one to recognize what is decisive and guide action. | 7 A GREAT MILITARY LEADER MAY BE THE CAPACITY NOT TO ACT IN THE ACTION AND TO ACT ON THE RESULTS. |
Military genius. Discernment allows one to choose when and whether to act, including when action is inadvisable. | 8 MILITARY GENIUS MAY BE THE CAPACITY NOT TO ACT. |
Military genius. Experience and intuition help a commander decide under uncertainty and pressure. | 9 MILITARY INTUITION AND EXPERIENCE MAY INTRODUCE INTO THE DECISION CHAIN THE TIME THAT DETERMINES DEFEAT. |
Surprise. Surprise can create an advantage, but preparing military means encounters physical, temporal, and organizational obstacles. | 10 WHEN THE MEANS OF ATTACK ARE ALREADY PRESENT IN CIVILIAN DEVICES, DEPLOYING AN ARMY MAY MEAN FACING AN OPPONENT WHO ALREADY RESIDES WITHIN YOUR OWN TERRITORY. |
Conclusion of Block 02
1. THE PARADOX OF TRUTH ARRIVING TOO LATE
In AI WAR, automation may accelerate the execution of decisions faster than the verification of their consequences.
A technically correct decision may arrive too late. A countermeasure may become obsolete before its effectiveness can even be demonstrated.
Truth does not cease to be true. The time to act on it may already have passed.
2. THE FOG OF TRUTH
The fog of war may arise from deception or a lack of information. It may also emerge from an excess of accurate information that military command cannot prioritize in time.
To manage this volume, command may delegate the selection of relevant information to AI. By determining what reaches the commander, that selection already shapes the judgment that follows.
The commander may retain the power to decide — while losing the power to see what the machine has discarded.
3. COMMAND RETURNED TOO LATE
Who authorizes a decision matters. So does who defined, before the crisis, its limits, the information available, and the alternatives that could be considered.
A commander may remain responsible for a decision while no longer being able to regain effective control during a crisis.
If the commander participates in every operation, human intervention may introduce delays. If control is returned only when an anomaly occurs, there may be too little context and too little time left to act.
Returning command does not mean restoring the ability to command.
A deeper paradox emerges. The chain of command may preserve its hierarchy, its officers, and its formal authority while technology silently changes the conditions under which decisions are made.
Command may remain in the same hands — while the power to define what can be decided shifts elsewhere.
4. THE WAR THAT HAS ALREADY CHANGED
Digital and conventional warfare may coexist while their conditions evolve at different speeds. Conventional war may itself accelerate technological transformations that would otherwise take years.
A military force may continue refining its traditional capabilities while digital threats alter its communications, intelligence, and ability to respond.
Perhaps there is a strategic mistake more serious than losing a war — continuing to prepare correctly for the war that began, when the war being fought has already changed its nature.
c) Distribution of Power and Capacity to Resist
Clausewitz × AI WAR
Reading note: The Clausewitz column contains conceptual paraphrases, not verbatim quotations. The AI WAR maxims are original formulations under investigation. The two columns are independent: the rows do not establish direct equivalence or refutation between their statements.
CLAUSEWITZ — CLASSICAL REFERENCES | AI WAR — ORIGINAL MAXIMS |
Dispersion and economy of force. Distributing resources may preserve capabilities, but it requires coordination and must serve the main effort. | 1 IN LOW-COST, DISPERSED ATTACK SYSTEMS, CENTRALIZING COORDINATION MAY CREATE COSTS AND VULNERABILITIES THAT DID NOT EXIST BEFORE. |
Defense and the culminating point. Defense seeks to preserve strength; prolonged efforts may consume the capacity needed to continue. | 2 EXPANDING DEFENSE MAY BE MORE COSTLY THAN ABANDONING IT. |
Culminating point. Prolonging an offensive may consume the superiority that made it possible to launch. | 3 SYSTEMS WITH LOW SUPPLY-CHAIN COSTS MAY IMPROVE AS THEY LEARN AND ADVANCE. THE COST OF CREATING THE SYSTEM MAY BE HIGH, BUT THE COST OF OPERATING IT IN AN ATTACK MAY BE LOW. |
Center of gravity. Identifying the decisive sources of power and cohesion guides strategy. | 4 IN A SWARM, NOT KNOWING THE ENEMY'S WEAK POINT DOES NOT PREVENT VICTORY. |
Numerical superiority. Numbers provide an important advantage, especially when they can be employed at the decisive point, but they do not guarantee victory. | 5 ATTACK AND CONFUSE: QUANTITY WORKS BY DISPERSING COUNTERMEASURES, WHILE A SINGLE ACCURATE STRIKE BRINGS IT DOWN. ACCURACY MAY BE PLANNED OR, BETTER YET, RANDOM. |
Attack and the culminating point. Attack seeks a positive gain, but prolonging it may consume the advantage that made it possible to begin. | 6 THE GAIN FROM AN ATTACK IS ALREADY AN END IN ITSELF. IT MAY COST LESS TO LAUNCH ANOTHER ATTACK USING WHAT WAS LEARNED THAN TO SUSTAIN THE FIRST. |
Culminating point and material superiority. A force advantage may be consumed in its use; superiority is neither absolute nor permanent. | 7 DENYING ACCESS TO THE BRUTE FORCE OF SILICON MAY FORCE AN ADVERSARY TO DISCOVER MORE EFFICIENT ARRANGEMENTS THAT REDUCE THE VALUE OF THE MONOPOLY. WHAT IS GAINED THROUGH DELAY IS LOST THROUGH IGNORANCE OF NEW TECHNOLOGICAL DEVELOPMENTS. |
Center of gravity and superiority. The sources of the whole's power and cohesion are fundamental; the quantity of available means is only one dimension of strength. | 8 INCREASINGLY, IT IS NOT THE BEST MODEL THAT CAUSES THE DAMAGE, BUT THE SMALLER ONE. |
Conclusion of Block 03
1. The Paradox of Tactical Success
A defender may intercept every attack and still fall behind in replenishing its resources.
Low-cost attacks can demand expensive responses at a pace faster than the defender can replenish its resources. Moreover, what is learned from one attack may reduce the cost of the next.
The advantage depends on how much it costs and how long it takes each side to replace what it consumes. Copying software, manufacturing drones and chips, securing energy supplies, and rebuilding industrial capacity involve different costs and timelines.
A defense can win every engagement and still move toward exhaustion.
2. The Paradox of Precision
When attacking becomes sufficiently cheap, volume can, under certain conditions, compensate for a lack of prior knowledge about where the enemy is vulnerable. Saturation can disperse defenses and create opportunities.
Hardened targets, resilient systems, and inexpensive countermeasures can reverse the advantage of volume. A target may be vulnerable yet have little significance for the military or political outcome.
Quantity may find a target. It does not necessarily find victory.
3. The Paradox of Control
An attacker may benefit from disorder. The defender must keep infrastructure, logistics, and command coordinated.
Excessive centralization creates vulnerable points. When central coordination fails, parts of a dispersed system may continue operating. Excessive decentralization can undermine the cohesion necessary to resist.
Beyond how much control to exercise, the question becomes what must remain connected and what must be able to survive disconnection.
Dispersal can preserve the parts. Integration must preserve the whole.
4. The Paradox of Technological Superiority
The capacity to create technology, the capacity to disrupt, and the capacity to sustain and restore operations under pressure are distinct dimensions of technological superiority.
Creating the most advanced intelligence may require an enormous concentration of capital, energy, and industrial capacity. Causing damage may require only a fraction of those resources. A smaller model may be sufficient for a task that does not require the most advanced technology.
Restricting an adversary's access to cutting-edge technology may buy time, but it can also encourage alternative architectures and more efficient methods, potentially reducing visibility into the adversary's technological development.
A state may expand its capabilities through externally controlled AI systems. If it depends on continued access to them, it becomes vulnerable to losing that access.
The power to build the technological frontier is not necessarily the power to resist.
Final Reflection — The Blind Spot of Military Readiness
A blackout awakens a state. A system that keeps running may conceal its own deterioration.
Attacks are repelled. Specialists are right within their domains. The chain of command retains its authority. Technology advances.
Everything appears to signal readiness — even as the time available to verify threats, coordinate decisions, replenish resources, and act grows shorter.
Technological evolution may make the same attack more damaging tomorrow — and harder to defend against than yesterday.
AI WAR does not render conventional armies irrelevant. It questions the very yardstick used to measure whether a state is prepared.
What if every indicator of readiness is correct — yet the state itself is no longer ready?
CONTINUE THE INVESTIGATION — AI WAR & TMC
AI WAR continues in two further essays:
2/3 — What If Defense Creates the Danger?
3/3 — Who Can Still Act When Everyone Attacks?
Both forthcoming.
The theoretical connection — TMC
The Theory of Continuous Mind (TMC) investigates mind, intelligence, and causal continuity, extending its framework to AI, AGI, and the prospect of superintelligence.
Chapter 09 — AGI (x): What defines general intelligence, and how might its development transform the system itself?
Chapter 10 — AGI f(x): Apparatus: A proposed computational architecture for AGI research, not an experimentally validated AGI system.
Related investigations
Who Can Still Surprise Claude? — Can AI systems building their successors recognize the blind spots they inherit?
The AGI (x) Effect: We Will Never Reach AGI. — Are we racing toward AGI without agreeing on what we are trying to achieve?






Comments