A Comprehensive Strategic, Technical, Historical, and Ethical Analysis (2017-2026)
 |
| Project Maven |
+Prologue: The Silent Transformation of War
War has always evolved with technology—from steel to gunpowder, from mechanization to nuclear deterrence. Yet no transformation has been as subtle, pervasive, and consequential as the one now unfolding: the transition from human decision-making in warfare to algorithmic mediation of violence.
At the center of this transformation lies Project Maven. What began as a technical solution to an intelligence bottleneck has evolved into something far more consequential: A system that compresses human judgment into machine-speed decision loops—reshaping not only how wars are fought, but how life-and-death decisions are made.
This documentary report examines Project Maven from its origins in 2017 through its global deployment in 2026. It traces the technical evolution, corporate partnerships, battlefield applications, and ethical implications of what has become the foundational architecture for algorithmic warfare in the 21st century.
Part I: Genesis—The Problem of Too Much Data (2017)
 |
| Transforming Data into Action |
1.1 The Intelligence Crisis
By 2016, U.S. military operations faced a paradox: unprecedented surveillance capability coupled with near-total inability to process collected data. Drone platforms alone generated millions of hours of Full Motion Video (FMV) and continuous ISR (Intelligence, Surveillance, and Reconnaissance) streams across multiple theaters.
The scale of the data challenge was staggering:
|
Metric |
Value |
Impact |
|
FMV
Generated Daily |
Thousands
of hours |
Overwhelming
volume |
|
Human
Review Rate |
< 5%
of data |
95%
unanalyzed |
|
Analyst
Burnout |
High
turnover |
Critical
skill loss |
|
Target
Detection Delay |
Hours to
days |
Missed
opportunities |
Table 1: The Intelligence Data Crisis (2016)
Human analysts could review less than 5% of collected intelligence. This created a strategic blind spot: critical threats existed inside data that no human would ever see. The problem was not a shortage of sensors or firepower, but the need to connect information and act faster than adversaries could respond.
1.2 The Birth of Project Maven
In April 2017, the Pentagon established the Algorithmic Warfare Cross-Functional Team (AWCFT), codenamed Project Maven. Lieutenant General John N.T. "Jack" Shanahan, then Director for Defense Intelligence (Warfighter Support), was appointed to lead this groundbreaking initiative. The mission was clear: automate the identification of objects and patterns in drone footage using machine learning.
The project was established in a memo by the U.S. Deputy Secretary of Defense on April 26, 2017, proposing an "Algorithmic Warfare Cross-Functional Team." With the help of the Defense Innovation Unit, the project obtained the support of top AI talents outside of the traditional defense contracting base. It was initially funded for $70 million over 36 months through rapid acquisition authorities.
Shanahan's vision was revolutionary. He pioneered the Department of Defense's first operational AI program, advancing the use of artificial intelligence for military operations and intelligence collection and analysis. According to Shanahan in November 2017, Maven was "designed to be that pilot project, that pathfinder, that spark that kindles the flame front of artificial intelligence across the rest of the [Defense] Department."
Initial Framing vs. Actual Trajectory:
- Initial: "Assist analysts" and "Reduce workload"
- Actual: Replace core elements of human cognitive labor in targeting
1.3 The Google Employee Revolt
When the Department of Defense first explored AI for
military use in 2017, its focus was highly specific: reduce the cognitive
burden of human drone pilots conducting search-and-kill missions against Middle
Eastern insurgents by automating the task of searching through video footage
for signs of enemy hideouts. To accomplish this mission, the Pentagon turned to
Google to generate the required software.
In 2018, the relationship between Google and the Pentagon
became a flashpoint for one of the most significant ethical confrontations in
Silicon Valley history. When thousands of Google employees signed a petition
opposing the company's involvement in a military-oriented project of this sort,
the company's leadership chose to terminate its contract for Maven.
The Google employee protest marked the first major ethical confrontation between Silicon Valley and military AI. It raised fundamental questions about the role of technology companies in warfare and the moral responsibilities of engineers building systems that could be used to take human lives.
Following Google's withdrawal, Shanahan reassigned the work
to Palantir, a defense-oriented startup chaired by Peter Thiel. Palantir then
developed the algorithms that enabled Maven software to identify potential
targets for attack by armed Predator drones. This transition marked a
fundamental shift in the corporate landscape of military AI.
Corporate Partnership Evolution:
|
Year |
Company |
Role |
Outcome |
|
2017-2018 |
Google |
AI
provider |
Withdrawal
(employee protest) |
|
2018-2024 |
Palantir |
Core
development |
MSS
platform deployed |
|
2024-2026 |
Multiple
vendors |
LLM
integration |
Decision-shaping
AI |
Table 2: Corporate Partnership Evolution
Part II: The Machine—Technical Architecture of Maven
The initial system capabilities focused on three core areas
that would transform military intelligence processing:
•
Object Detection: Humans,
vehicles, weapons
•
Pattern Recognition:
Movement, formation behavior
•
Activity Classification:
Suspicious vs. normal patterns
This phase addressed what military planners called the
"visibility problem"—making vast quantities of collected data
readable and actionable. The output included annotated video feeds, highlighted
targets, and analyst alerts that dramatically accelerated the intelligence
processing pipeline.
As of December 2017, 150,000 images had been manually
labeled to establish the first training data sets, with projections to reach 1
million by January 2018. This massive data labeling effort was essential for
training the machine learning algorithms that would power the system's
recognition capabilities.
2.2 Phase Two: From Tool to Platform
Maven evolved into the Maven Smart System (MSS)—a
comprehensive battlefield intelligence platform with four key components that
would revolutionize military decision-making:
MSS Architecture Components:
|
Component |
Function |
Capability |
|
Data
Fusion Engine |
Multi-source
integration |
Satellite,
drone, SIGINT, OSINT |
|
Ontology
Layer |
Structuring
reality |
Object →
Event → Relationship → Context |
|
NLP
Command Layer |
Natural
language queries |
LLM-based
intelligence synthesis |
|
AI
Tasking System |
Strike
recommendation |
Target
prioritization, weapon selection |
Table 3: Maven Smart System Architecture
The most significant evolution was the shift from merely
detecting targets to recommending how to eliminate them. System outputs now
include target prioritization, strike options, weapon selection, and timing
optimization—transforming AI from an analytical tool into a decision-support
system with lethal implications.
2.3 GEOINT and the NGA's Role
The National Geospatial-Intelligence Agency (NGA) plays a
decisive role in U.S. national security by providing geospatial intelligence
(GEOINT) for military, policy, and disaster response needs. At the GEOINT
Symposium of 2022, it was announced that Project Maven was transferred from the
Office of the Under Secretary of Defense for Intelligence and Security to the
NGA, under President Biden's proposed budget for Fiscal Year 2023. It became a
Program of Record on November 7, 2023.
NGA's state-of-the-art computer vision and AI capabilities
are now integrated into various military analytic workflows to automatically
detect, identify, characterize, extract, and attribute features and objects in
imagery and video. Maven provides trusted GEOINT at speed and scale for object
recognition.
NGA Maven has decreased targeting workflow timelines by a substantial amount, with one of our fighting element's targeting cells seeing intelligence operation timelines drop from hours to minutes—from sensing to target engagement—during a recent exercise.
According to NGA Director Vice Admiral Frank Whitworth, NGA
Maven is now available to all services and all combatant commands, with 20,000
active users through more than 35 service and combat and command tools across
three security domains. The user base has more than quadrupled since March of
last year.
NGA Maven Operational Impact:
|
Metric |
Traditional |
Maven-Enabled |
|
Targeting
Timeline |
Hours |
Minutes |
|
Active
Users |
Hundreds |
20,000+ |
|
Combatant
Commands |
Limited |
All
commands |
|
Decision
Quality |
Variable |
1,000/hour
target capacity |
Table 4: NGA Maven Performance Metrics (2025)
Part III: The Doctrinal Shift—From Kill Chain to Kill Web
3.1 From Kill Chain to Kill Web
The traditional targeting process followed a linear sequence
that reflected industrial-era military thinking:
- Sensor detects
- Analyst evaluates
- Commander decides
- Weapon deployed
The Maven-enabled model transforms this into a networked
system: multiple sensors feeding AI fusion layer, automated prioritization, and
distributed execution. This represents a fundamental shift from sequential to
parallel processing of targeting decisions.
Decision Timeline Comparison:
|
Phase |
Traditional |
Maven-Enabled |
Compression |
|
Detection
to Analysis |
Hours |
Minutes |
60x
faster |
|
Analysis
to Decision |
Hours |
Seconds |
100x+
faster |
|
Decision
to Strike |
Variable |
Automated |
Near-instant |
Table 5: Decision Loop Compression
3.2 CJADC2 and All-Domain Operations
Combined Joint All-Domain Command and Control (CJADC2) was
created to operate in the reality of modern warfare, where actions in the air
can trigger effects in space, cyberspace, or at sea within seconds. CJADC2 is
the U.S. Department of Defense's evolving framework for enabling faster, more
effective decision-making by linking sensors, commanders, and shooters across
services and coalition partners.
CJADC2 aims to: collect data from sensors across all
domains; process and fuse that data into a coherent operational picture;
distribute actionable information to the right decision-makers and weapons
systems; and enable rapid, synchronized action across services and allies.
Rather than a single system or platform, CJADC2 is a concept, architecture, and
approach for how future forces share data, coordinate actions, and fight as an
integrated whole.
In a CJADC2-enabled environment, a sensor detecting a threat
in one domain—such as a Space-Based Infrared System satellite tracking a
missile launch, an Aegis-equipped Navy destroyer radar identifying a hostile
aircraft, or an Army Sentinel radar spotting incoming rockets—can immediately
share that data across the force. This approach decouples sensors from
shooters, allowing the most suitable platform and weapon to respond regardless
of service or domain.
The Pentagon aims to utilize AI tools like Maven to support
its CJADC2 warfighting construct. This initiative seeks to better connect the
platforms, sensors, and data streams of the U.S. military and its key
international partners under a unified network. Defense officials believe that
leveraging AI will help commanders and other personnel make faster and more
informed decisions, thereby improving operational effectiveness and efficiency.
Part IV: The Corporate-Military Complex
4.1 Rise of Palantir
Following Google's withdrawal from Project Maven in 2018,
Palantir took over core system development, building the Maven Smart System
with focus on data integration, battlefield ontology, and operational
scalability. Under the leadership of CEO Alex Karp and Chairman Peter Thiel,
Palantir has become the dominant player in military AI infrastructure.
Palantir's commercialized Maven Smart System pulls together
data of different classification levels from a vast array of sources—satellite
intelligence on potential enemy targets, readiness reports from friendly units,
social media posts on unfolding crises or misinformation—and puts it into a
single, customizable interface for military planners.
On May 29, 2024, Palantir was awarded a $480 million
contract by the U.S. Army for its Maven Smart System prototype. This five-year
firm-fixed-price contract, running through May 28, 2029, will allow the Defense
Department to expand its use to thousands of users at five combatant commands:
U.S. Central Command, European Command, Indo-Pacific Command, Northern Command,
and Transportation Command. The system will also be available to members of the
Joint Staff.
Palantir Defense Contracts (2024-2026):
|
Contract |
Value |
Purpose |
|
Army MSS
Expansion |
$480
million |
Expand to
5 combatant commands |
|
Army
Research Lab |
$100
million |
MSS
support for all services |
|
Army
Licenses (May 2025) |
$795
million |
New MSS
licenses |
|
NGA
Expansion |
$28
million |
MSS for
NGA analysts |
Table 6: Palantir Major Defense Contracts
According to Shannon Clark, Palantir's head of defense
growth, "Users are going to span everyone from intel analysts and
operators in some of the remote island chains across the world to leadership at
the Pentagon. This is taking what has been built in prototype and
experimentation and bringing this to production."
4.2 NATO Adoption
In April 2025, NATO announced it had awarded a contract to
Palantir to adopt its Maven Smart System for artificial intelligence-enabled
battlefield operations. Through the contract, finalized March 25, the NATO
Communications and Information Agency (NCIA) plans to use Maven Smart System
NATO to support the transatlantic military organization's Allied Command
Operations strategic command.
NATO plans to use the system to provide a common
data-enabled warfighting capability to the Alliance, through a wide range of AI
applications—from large language models (LLMs) to generative and machine
learning—ultimately enhancing intelligence fusion and targeting, battlespace
awareness and planning, and accelerated decision-making.
The contract was one of the most expeditious in NATO's history, taking only six months from outlining the requirement to acquiring the system.
Ludwig Decamps, NCIA General Manager, stated that the deal
with Palantir is focused on "providing customized state-of-the-art AI
capabilities to the Alliance, and empowering our forces with the tools required
on the modern battlefield to operate effectively and decisively."
4.3 The Anthropic Dispute
In 2025, Anthropic mustered Claude, its large language
model, for national service. Although the military-industrial complex is newly
fashionable, Anthropic was not a natural fit. The firm had been founded in 2021
by seven OpenAI defectors who believed that AI safety should be prioritized.
The company's CEO, Dario Amodei, wanted Claude to be helpful at the most
sensitive level—Claude was the first AI certified to operate on classified
systems.
The Pentagon has been using Claude to analyze data, write
memos, and help generate battle plans. Intelligence contractors like Palantir
offer platforms that synthesize, process, and surface decision-relevant
information. As one Palantir employee noted, "Claude is just the best, by
far." A human analyst might review signal intelligence to select military
targets; Claude can do the same thing, only much faster and more efficiently.
However, tensions emerged when the Pentagon sought to
renegotiate the contract to include "all lawful uses" of the product.
Anthropic had stipulated that Claude be used neither to drive fully autonomous
weaponry nor to facilitate domestic mass surveillance. The Pentagon accepted
these stipulations initially, but later sought to remove them.
On February 27, 2025, Defense Secretary Pete Hegseth officially declared Anthropic a supply-chain risk, stating that "no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic." This designation, which had only ever been applied to infrastructure firms with ties to adversarial foreign governments, threatened to extinguish the company.
Anthropic filed two lawsuits challenging the
constitutionality of the ban. The company maintains that it cannot manipulate
Claude once deployed—there is no remote kill switch, no backdoor, and no
mechanism to push unauthorized updates. The dispute represents a fundamental
clash between Silicon Valley's ethical AI movement and the Pentagon's desire
for unrestricted use of AI capabilities.
Part V: Real-World Deployment—From Theory to Practice
5.1 Gaza: AI-Driven Targeting Ecosystem
Gaza represents the most controversial and extensively
documented deployment environment for AI targeting systems. Israeli forces have
relied heavily on multiple AI tools that together constitute an automation of
the find-fix-track-target components of the modern military "kill
chain."
The Three Core AI Systems:
The Gospel (Habsora):An
AI-powered database that generates targets based on apparent links to Hamas.
During Israel's 11-day war with Hamas in May 2021, The Gospel generated 100
targets daily—a significant increase from the previous rate of 50 targets per
year in Gaza. The system was developed by Unit 8200, Israel's elite
intelligence and cyber technology unit.
Lavender:An AI
recommendation system designed to use algorithms to identify Hamas operatives
as targets. Lavender scans information on approximately 90% of Gaza's
population and gives each individual a rating between 1 to 100, expressing the
likelihood that the individual is a member of Hamas or Islamic Jihad military
wings.
Where's Daddy?:A
grotesquely named system that tracks targets geographically so they can be
followed into their family residences before being attacked. One intelligence
officer told +972 Magazine: "We were not interested in killing operatives
only when they were in a military building or engaged in a military activity.
On the contrary, the IDF bombed them in homes without hesitation, as a first
option. It's much easier to bomb a family's home."
AI Targeting Systems in Gaza:
|
System |
Function |
Scale |
|
The
Gospel |
Infrastructure
targeting |
100
targets/day (vs. 50/year pre-AI) |
|
Lavender |
Individual
targeting |
37,000+
people rated |
|
Where's
Daddy? |
Home
location tracking |
Family
residence targeting |
Table 7: Israeli AI Targeting Systems in Gaza
According to intelligence officers who spoke with +972
Magazine, sources revealed that approximately 10% of the people that Lavender
marked to be killed were not Hamas militants—some had loose connections to
Hamas, while others had completely no connection. The machine would bring
people who had the exact same name and nickname as a Hamas operative, or people
who had similar communication profiles, including civil defense workers and
police officers in Gaza.
One source said he spent only 20 seconds per target before authorizing the bombing of alleged low-ranking Hamas militants—often civilians—killing those people inside their houses.
The combination of Lavender and Where's Daddy? led to entire
Palestinian families being wiped out inside their houses. According to U.N.
statistics, more than 50% of casualties in the first six weeks came from a
smaller group of families—an expression of the family unit being destroyed by
AI-enabled targeting.
5.2 Iran 2026: The Minab School Tragedy
On February 28, 2026, the first day of the Iran war, the
Shajareh Tayyebeh girls' elementary school in Minab, Hormozgan province,
southern Iran, was destroyed by missile strikes. According to witness accounts
verified by satellite-based analyses, the school was triple-tapped by three
distinct strikes. The roof collapsed on students, and according to Iranian
media, between 175 and 180 people were killed, most of whom were
schoolchildren.
Human rights organization Hengaw stated that around 170
students were present in the school at the time, while the Iranian Ministry of
Education said 264 students were present, mostly girls between seven and 12
years old. The impact instantaneously killed dozens inside, destroying at least
half of the two-story school building.
According to testimony from Red Crescent medics and victims'
parents, the initial strike was followed by a "double-tap" strike.
The school's principal moved students to a prayer room and called parents; that
area was then hit by a second strike, killing most who had taken shelter.
According to Minab's mayor, the school was triple-tapped—struck three times in
total.
Investigations by The New York Times, CBC, NPR, BBC Verify, and others concluded that the United States was likely responsible for the strike. Sources involved with the US military's internal investigation corroborated that the strike was likely perpetrated by the US, despite the investigation not yet having reached a final conclusion.
The root cause was misclassification by the AI system
combined with reliance on outdated targeting data. The New York Times reported
that the U.S. preliminary investigation found that the United States is
responsible for this strike due to outdated targeting data. This incident
exemplifies the fundamental risk of algorithmic warfare: at scale, errors are
not isolated incidents but become systemic outcomes with devastating
humanitarian consequences.
On March 13, 2026, Congressman Jason Crow and 120 members of
Congress demanded answers on the school strike in Iran, writing to Secretary
Hegseth to express alarm regarding reports of civilian casualties arising from
Operation Epic Fury. The letter specifically cited the Minab school attack as
"the deadliest attack on civilians thus far" and demanded clear
answers on how the Department plans to investigate these reports and prevent
the risk of further civilian harm.
Part VI: The Ethical Collapse Point
6.1 Automation Bias and Human Control
The integration of AI into targeting has created multiple
ethical crisis points. Automation bias—the tendency of humans to trust machine
outputs and reduce independent verification—creates a dangerous dependency on
algorithmic judgment. AI enables thousands of decisions per hour; humans cannot
validate at that speed, creating an inherent tension between operational
efficiency and moral accountability.
Contrary to popular belief, the Department of Defense has
never had a policy requiring autonomous weapons to have a "human in the
loop." What DoD Directive 3000.09 states is that autonomous weapons
"will be designed to allow commanders and operators to exercise
appropriate levels of human judgment over the use of force." This subtle
but crucial distinction has significant implications for how AI warfare is
actually conducted.
The problem with the "human-in-the-loop" framing is that it presumes a machine decision loop and asks where the human is relative to that pre-existing loop. Rather than give primacy to the machine's work, we should prioritize and make the human's decision cycle central.
Research from the Hellenic Air Force Academy's War Games Lab
found that operators who took AI suggestions into consideration made decisions
aligned with International Humanitarian Law and Rules of Engagement 78% of the
time. However, 36% of operators discussed the risk of over-trusting AI's
suggestions due to time constraints, and 88% emphasized the need for constant
training on the platform as well as on ethics and legal constraints.
6.2 Accountability Vacuum
There is no clear responsibility between engineers,
commanders, and AI systems when errors occur. The transformation from limited,
deliberate strikes to continuous target production pipelines represents a
fundamental change in the nature of warfare itself. Targets become data points
and probability scores rather than human lives, fundamentally altering the
psychological and moral framework of warfare.
AI-enabled targeting systems, even those that retain
humans-in-the-loop, generate significant moral challenges. Such systems draw on
vast volumes of data, virtually guaranteeing an opaque process of data
crunching, analyzing, and target proposition. Human operators are unlikely to
have a clear overview of what data such systems have available, what they are
trained on, what the specific parameters are for algorithmic calculations, or
what frequent updates do to accuracy.
Summary of
Ethical Concerns:
|
Issue |
Description |
Severity |
|
Automation
Bias |
Over-reliance
on AI outputs |
High |
|
Speed vs.
Ethics |
Cannot
validate at machine speed |
Critical |
|
Dehumanization |
Targets
as data points |
Critical |
|
Accountability
Gap |
Unclear
responsibility chain |
High |
|
Violence
Industrialization |
Continuous
targeting pipeline |
Critical |
Table 8: Ethical Concerns Summary
Part VII: Strategic Consequences
7.1 The Global AI Arms Race
A military artificial intelligence arms race has emerged
between major powers to develop and deploy advanced AI technologies and lethal
autonomous weapons systems. The goal is to gain strategic or tactical advantage
over rivals, similar to previous arms races involving nuclear or conventional
military technologies.
Russian President Vladimir Putin stated that the leader in
AI will "rule the world." An AI arms race is sometimes placed in the
context of an AI Cold War between the United States and China. Researchers warn
that the AGI race between major powers could reshape geopolitical power,
including AI for surveillance, autonomous weapons, decision-making systems, and
cyber operations.
The competitive pressure to automate creates a destabilizing
dynamic. As one nation deploys AI-enabled targeting systems, others feel
compelled to follow suit, potentially leading to a destabilizing arms race in
autonomous weapons systems. The ethical implications extend beyond the
battlefield: when targeting becomes faster than verification and larger than
human oversight, errors become structural outcomes rather than isolated
accidents.
AI Arms Race Participants:
|
Nation |
AI Military Focus |
Key Programs |
|
United
States |
Decision
support, targeting |
Maven,
CJADC2, JADC2 |
|
China |
Surveillance,
autonomous systems |
AI-enabled
ISR, drone swarms |
|
Russia |
Autonomous
weapons, cyber |
Lethal
autonomous systems |
|
Israel |
Target
identification |
Lavender,
Gospel, Where's Daddy |
Table 9: Global AI Military Development
7.2 Transformation of Military Roles
The human role shifts from decision-maker to system
supervisor. This represents a fundamental redefinition of military
professionalism and command responsibility. AI-enabled warfare reduces time for
diplomacy and increases escalation risk. The compression of decision timelines
leaves less room for de-escalation and negotiation.
Rapid adoption by major powers creates competitive pressure
to automate, potentially leading to a destabilizing arms race in autonomous
weapons systems. The ethical implications extend beyond the battlefield: when
targeting becomes faster than verification and larger than human oversight,
errors become structural outcomes rather than isolated accidents.
The transformation affects every level of military
operations. Intelligence analysts now work alongside AI systems that can
process data thousands of times faster than humans. Commanders must make
decisions based on AI-generated recommendations with limited time for
independent verification. The traditional skills of military judgment,
situational awareness, and ethical reasoning are being supplemented—and in some
cases replaced—by algorithmic decision-making.
Part VIII: The Future of War
The emerging reality includes AI-integrated battle networks,
autonomous targeting assistance, and global real-time surveillance. The next
phase will feature AI-coordinated warfare ecosystems where multiple systems
operate in concert. The trajectory is clear: war is no longer constrained by
human limits, but by algorithmic capability.
The integration of AI in military operations aims to enhance
the speed and accuracy of target identification. Positive target identification
(PID) is at the forefront of the targeting process. The speed at which a
hostile target can be detected is crucial to the remaining steps of the
targeting cycle (Decide, Detect, Deliver, Assess). AI assists by filtering
specific user-defined parameters, sifting through large amounts of data,
extracting what is relevant, and providing analysts with near-real-time data used
by the operations community for validation against the commander's objective.
Emerging Capabilities Roadmap:
|
Capability |
Status |
Timeline |
|
AI-Integrated
Battle Networks |
Deployed |
2024-2026 |
|
Autonomous
Targeting Assistance |
Operational |
2025-2027 |
|
Global
Real-Time Surveillance |
In
Development |
2026-2028 |
|
AI-Coordinated
Warfare Ecosystems |
Emerging |
2027-2030 |
Table 10: Future Warfare Capabilities Roadmap
The question is no longer whether AI will transform warfare.
It already has. The question is whether humanity can maintain meaningful
control over the machines we have created to kill on our behalf. As one
military ethicist observed: ethics is not a constraint on military
operations—it is a force multiplier. Targeting decisions that are legally
grounded, morally defensible, and procedurally transparent ensure operational
legitimacy, effectiveness, and public trust.
Looking ahead, the international community must grapple with
urgent questions: How do we regulate algorithmic warfare? What constraints
should be placed on AI targeting systems? How do we ensure accountability when
machines make lethal decisions? The answers to these questions will determine
whether the future of war remains a human endeavor—or becomes something else
entirely.
Final Conclusion: The Industrialization of Decision and Death
Project Maven represents a fundamental shift in warfare. It
solved the problem of too much data and created a system capable of
machine-speed targeting. However, it introduced a new risk: decision-making
detached from human cognition.
War is no longer constrained by human limits. It is
constrained by algorithmic capability.
The ultimate ethical reality is stark: When targeting
becomes faster than verification and larger than human oversight, errors are no
longer accidents—they become structural outcomes. The Minab school tragedy, the
Gaza targeting operations, and countless other incidents demonstrate this
fundamental truth.
The debate about "human in the loop" misses the
essential point. What matters is not the presence of humans in the process, but
the preservation of human judgment, moral responsibility, and accountability.
When AI systems generate thousands of targets per day, when human review is
reduced to seconds per target, when family homes are bombed because an
algorithm calculated it was "easier" than targeting militants in
military contexts—the loop has already been broken.
The question is no longer whether AI will
transform warfare. It already has. The question is whether humanity can
maintain meaningful control over the machines we have created to kill on our
behalf. Project Maven has given us the answer: not without a fundamental
recommitment to human judgment, ethical constraints, and the recognition that
efficiency in killing is not the same as justice in war.
As military AI continues to
evolve, the international community must grapple with urgent questions: How do
we regulate algorithmic warfare? What constraints should be placed on AI
targeting systems? How do we ensure accountability when machines make lethal
decisions? The answers to these questions will determine whether the future of
war remains a human endeavor—or becomes something else entirely.
The legacy of Project Maven extends far beyond its technical
achievements. It has established the template for algorithmic warfare in the
21st century—a template that is being adopted by nations around the world. The
choices we make now about how to govern these systems will shape the nature of
conflict for generations to come.
Resources:
https://mwi.westpoint.edu/big-data-at-war-special-operations-forces-project-maven-and-twenty-first-century-warfare/
https://www.researchgate.net/publication/399487363_Algorithmic_Warfare_in_Evolution_A_Cyber-Conflict_Analysis_of_Israel's_AI_Targeting_from_Gaza_2021_to_Gaza_2023-2025
https://www.independent.co.uk/news/world/americas/project-maven-ai-us-airstrike-iraq-anthropic-b2929138.html
https://www.nga.mil/news/GEOINT_Artificial_Intelligence_.html
https://debuglies.com/2025/12/17/beyond-tactical-brilliance-the-decadal-shift-to-human-machine-fusion/
https://www.researchgate.net/publication/392770698_Improving_Arabic_Image_Captioning_with_Vision-Language_Models
https://arxiv.org/html/2503.21910v1
https://aoav.org.uk/2025/kill-codes-and-command-lines-understanding-the-rise-of-algorithmic-warfare/
https://odsc.medium.com/palantir-secures-480-million-dod-deal-for-ai-powered-maven-smart-system-prototype-2869135cbc90
https://fedscoop.com/project-maven-dod-machine-learning/
https://www.army.mil/article/290021/data_centric_command_and_control_unlocking_mercurys_potential_with_c2_next
https://www.aflcmc.af.mil/NEWS/Article/4241273/air-force-battle-lab-advances-the-kill-chain-with-ai-c2-innovation/
https://blog.palantir.com/maven-smart-system-innovating-for-the-alliance-5ebc31709eea
https://www.c4isrnet.com/it-networks/2018/07/27/targeting-the-future-of-the-dods-controversial-project-maven-initiative/
https://www.digitaldividedata.com/blog/geospatial-data-geoint-use-cases-in-defense-tech
https://publications.armywarcollege.edu/News/Display/Article/4361748/mission-commands-asymmetric-advantage-through-ai-driven-data-management/
https://reliefweb.int/report/world/human-loop-how-oversight-turns-ai-humanitarian-ally
https://www.dvidshub.net/news/329789/sensor-shooter-faster
https://www.defenceiq.com/glossary/intelligence-surveillance-target-acquisition-and-reconnaissance
https://www.theguardian.com/technology/2026/mar/13/anthropic-pentagon-artificial-intelligence
https://www.tandfonline.com/doi/full/10.1080/16544951.2025.2540131
https://www.researchgate.net/publication/389355214_AI_Translation_of_the_Gaza-Israel_War_Terminology
https://v45.diplomacy.edu/updates/anthropic-pentagon-military-ai
https://www.war.gov/News/News-Stories/article/article/1356172/project-maven-industry-day-pursues-artificial-intelligence-for-dod-challenges/
https://www.esd.whs.mil/Portals/54/Documents/FOID/Reading%20Room/Other/15-F-0070_DOC_05_FINAL_ODNA_Revolution_in_Military_Affairs_Conference_November_1996.pdf
https://sites.bu.edu/pardeeatlas/research-and-policy/the-effectiveness-of-employee-activism-in-big-tech-dod-partnerships/
https://ndupress.ndu.edu/Media/News/News-Article-View/Article/2054156/the-ethics-of-acquiring-disruptive-technologies-artificial-intelligence-autonom/
https://www.nga.mil/assets/files/170901-038_GEOINT_Basic_Doctrine_Pub_1.pdf
https://www.lawfaremedia.org/article/military-ai-policy-by-contract--the-limits-of-procurement-as-governance
https://www.missiledefenseadvocacy.org/maven-smart-system/
https://www.orfonline.org/english/expert-speak/ai-in-real-time-warfare-lessons-from-project-maven/
https://www.noonpost.com/361260/
https://www.maris-tech.com/blog/what-is-istar-intelligence-surveillance-target-acquisition-and-reconnaissance/
https://lieber.westpoint.edu/ai-based-targeting-gaza-surveying-expert-responses-refining-debate/
https://www.jns.org/behind-the-scenes-of-the-idfs-war-drill-a-digital-revolution/
https://organiser.org/2026/03/09/343310/bharat/maven-smart-system-an-artificial-intelligence-that-is-shaping-us-iran-war-do-we-have-a-similar-system/
https://www.cnas.org/publications/commentary/project-maven-brings-ai-to-the-fight-against-isis
https://emerj.com/big-data-military/
https://www.mitchellaerospacepower.org/app/uploads/2021/02/a2dd91_4892807f169341188b7ebcd2f775671d.pdf
https://news.futunn.com/en/post/70095661/the-real-ai-for-warfare-claude-is-just-the-foundation
https://defensescoop.com/2025/04/14/nato-palantir-maven-smart-system-contract/
https://www.habtoorresearch.com/programmes/maduro-khamenei-artificial-intelligence/
https://journals.law.harvard.edu/nsj/2025/05/on-the-pitfalls-of-technophilic-reason-a-commentary-on-kevin-jon-hellers-the-concept-of-the-human-in-the-critique-of-autonomous-weapons/
https://www.reddit.com/r/ArtificialInteligence/comments/1rqpghu/project_maven_palantir_and_anthropic/
https://sundayguardianlive.com/world/how-israel-is-using-ai-in-its-wars-in-gaza-and-iran-174676/
https://www.faf.ae/home/2026/3/12/x1-1
https://www.brennancenter.org/our-work/research-reports/business-military-ai
https://baptistnews.com/article/ai-goes-to-war-and-schoolchildren-are-dead/
https://www.the-independent.com/news/world/americas/project-maven-ai-us-airstrike-iraq-anthropic-b2929138.html
https://systematic.com/int/industries/defence/products/deep-dives/fire-support/
https://breakingdefense.com/2019/11/secarmys-multi-domain-kill-chain-space-to-cloud-to-ai/
https://projectgeospatial.org/geospatial-frontiers/the-new-battlespace-how-geospatial-ai-is-reshaping-military-intelligence
https://aoav.org.uk/2026/who-commands-the-god-in-the-machine-ai-and-the-future-of-military-authority/
https://www.salesforce.com/blog/ai-and-human-touch/
https://politicalkeys.net/?p=6655
https://www.arabnews.com/node/2624225/amp
https://www.madarcenter.org/
https://en.wikipedia.org/wiki/Project_Maven
https://comptroller.war.gov/Portals/45/Documents/defbudget/fy2020/budget_justification/pdfs/01_Operation_and_Maintenance/O_M_VOL_1_PART_1/Volume_1_Part_1.pdf
https://mwi.westpoint.edu/big-data-at-war-special-operations-forces-project-maven-and-twenty-first-century-warfare/
https://politicalkeys.net/?p=6655
https://apps.dtic.mil/sti/tr/pdf/ADA353436.pdf
https://aclanthology.org/2020.alvr-1.1.pdf
https://www.researchgate.net/publication/399487363_Algorithmic_Warfare_in_Evolution_A_Cyber-Conflict_Analysis_of_Israel's_AI_Targeting_from_Gaza_2021_to_Gaza_2023-2025
https://www.independent.co.uk/news/world/americas/project-maven-ai-us-airstrike-iraq-anthropic-b2929138.html
https://api.army.mil/e2/c/downloads/361884.pdf
https://www.nga.mil/news/GEOINT_Artificial_Intelligence_.html
https://debuglies.com/2025/12/17/beyond-tactical-brilliance-the-decadal-shift-to-human-machine-fusion/
https://www.researchgate.net/publication/392770698_Improving_Arabic_Image_Captioning_with_Vision-Language_Models
https://arxiv.org/html/2503.21910v1
https://aoav.org.uk/2025/kill-codes-and-command-lines-understanding-the-rise-of-algorithmic-warfare/
https://en.wikipedia.org/wiki/Project_Maven
https://apps.dtic.mil/sti/pdfs/AD1179110.pdf
https://dsb.cto.mil/wp-content/uploads/reports/2000s/ADA463361.pdf
https://odsc.medium.com/palantir-secures-480-million-dod-deal-for-ai-powered-maven-smart-system-prototype-2869135cbc90
https://fedscoop.com/project-maven-dod-machine-learning/
https://www.army.mil/article/290021/data_centric_command_and_control_unlocking_mercurys_potential_with_c2_next
https://www.aflcmc.af.mil/NEWS/Article/4241273/air-force-battle-lab-advances-the-kill-chain-with-ai-c2-innovation/
https://blog.palantir.com/maven-smart-system-innovating-for-the-alliance-5ebc31709eea
https://www.c4isrnet.com/it-networks/2018/07/27/targeting-the-future-of-the-dods-controversial-project-maven-initiative/
https://www.digitaldividedata.com/blog/geospatial-data-geoint-use-cases-in-defense-tech
https://publications.armywarcollege.edu/News/Display/Article/4361748/mission-commands-asymmetric-advantage-through-ai-driven-data-management/
https://reliefweb.int/report/world/human-loop-how-oversight-turns-ai-humanitarian-ally
https://www.andrewwmarshallfoundation.org/wp-content/uploads/2022/11/AIRMA_FINAL.pdf
https://www.dvidshub.net/news/329789/sensor-shooter-faster
https://www.defenceiq.com/glossary/intelligence-surveillance-target-acquisition-and-reconnaissance
https://en.wikipedia.org/wiki/Glossary_of_military_abbreviations
https://www.theguardian.com/technology/2026/mar/13/anthropic-pentagon-artificial-intelligence
https://www.tandfonline.com/doi/full/10.1080/16544951.2025.2540131
https://www.researchgate.net/publication/389355214_AI_Translation_of_the_Gaza-Israel_War_Terminology
https://www.marines.mil/portals/1/mcrp%205-12a%20with%20ch.%201%20z.pdf
https://v45.diplomacy.edu/updates/anthropic-pentagon-military-ai
https://www.war.gov/News/News-Stories/article/article/1356172/project-maven-industry-day-pursues-artificial-intelligence-for-dod-challenges/
https://www.arabnews.com/node/2624225/amp
https://www.madarcenter.org/
https://www.esd.whs.mil/Portals/54/Documents/FOID/Reading%20Room/Other/15-F-0070_DOC_05_FINAL_ODNA_Revolution_in_Military_Affairs_Conference_November_1996.pdf
https://sites.bu.edu/pardeeatlas/research-and-policy/the-effectiveness-of-employee-activism-in-big-tech-dod-partnerships/
https://ndupress.ndu.edu/Media/News/News-Article-View/Article/2054156/the-ethics-of-acquiring-disruptive-technologies-artificial-intelligence-autonom/
https://www.nga.mil/assets/files/170901-038_GEOINT_Basic_Doctrine_Pub_1.pdf
https://www.lawfaremedia.org/article/military-ai-policy-by-contract--the-limits-of-procurement-as-governance
https://www.missiledefenseadvocacy.org/maven-smart-system/
https://www.orfonline.org/english/expert-speak/ai-in-real-time-warfare-lessons-from-project-maven/
https://www.noonpost.com/361260/
https://www.maris-tech.com/blog/what-is-istar-intelligence-surveillance-target-acquisition-and-reconnaissance/
https://lieber.westpoint.edu/ai-based-targeting-gaza-surveying-expert-responses-refining-debate/
https://www.jns.org/behind-the-scenes-of-the-idfs-war-drill-a-digital-revolution/
https://organiser.org/2026/03/09/343310/bharat/maven-smart-system-an-artificial-intelligence-that-is-shaping-us-iran-war-do-we-have-a-similar-system/
https://www.cnas.org/publications/commentary/project-maven-brings-ai-to-the-fight-against-isis
https://en.wikipedia.org/wiki/Intelligence,_surveillance,_target_acquisition,_and_reconnaissance
https://www.refaad.com/Files/JALLS/JALS-6-2-4.pdf
https://en.wikipedia.org/wiki/Anthropic%E2%80%93United_States_Department_of_Defense_dispute
https://emerj.com/big-data-military/
https://www.mitchellaerospacepower.org/app/uploads/2021/02/a2dd91_4892807f169341188b7ebcd2f775671d.pdf
https://news.futunn.com/en/post/70095661/the-real-ai-for-warfare-claude-is-just-the-foundation
https://defensescoop.com/2025/04/14/nato-palantir-maven-smart-system-contract/
https://irp.fas.org/doddir/army/adp1_02.pdf
https://www.habtoorresearch.com/programmes/maduro-khamenei-artificial-intelligence/
https://journals.law.harvard.edu/nsj/2025/05/on-the-pitfalls-of-technophilic-reason-a-commentary-on-kevin-jon-hellers-the-concept-of-the-human-in-the-critique-of-autonomous-weapons/
https://www.europarl.europa.eu/meetdocs/2004_2009/documents/dv/270/270907/270907nolingeneral_en.pdf
https://www.reddit.com/r/ArtificialInteligence/comments/1rqpghu/project_maven_palantir_and_anthropic/
https://www.youtube.com/watch?v=7QTPIY9wRDU
https://library.au.int/fr/modern-military-dictionary-english-arabic-arabic-english-4
https://www.usmcu.edu/portals/218/jams_fall2020_11_2_web2.pdf
https://sundayguardianlive.com/world/how-israel-is-using-ai-in-its-wars-in-gaza-and-iran-174676/
https://ieeexplore.ieee.org/iel8/6287639/10820123/11146650.pdf
https://www.faf.ae/home/2026/3/12/x1-1
https://comptroller.war.gov/Portals/45/Documents/defbudget/fy2020/budget_justification/pdfs/01_Operation_and_Maintenance/O_M_VOL_1_PART_1/Volume_1_Part_1.pdf
https://www.brennancenter.org/our-work/research-reports/business-military-ai
https://baptistnews.com/article/ai-goes-to-war-and-schoolchildren-are-dead/
https://www.salesforce.com/blog/ai-and-human-touch/
https://www.the-independent.com/news/world/americas/project-maven-ai-us-airstrike-iraq-anthropic-b2929138.html
https://systematic.com/int/industries/defence/products/deep-dives/fire-support/
https://breakingdefense.com/2019/11/secarmys-multi-domain-kill-chain-space-to-cloud-to-ai/
https://projectgeospatial.org/geospatial-frontiers/the-new-battlespace-how-geospatial-ai-is-reshaping-military-intelligence
https://unidir.org/wp-content/uploads/2023/05/Table-Top-Exercises-on-the-Human-Element-and-Autonomous-Weapons-Systems-Summary-Report-UNIDIR-Final.pdf
https://aoav.org.uk/2026/who-commands-the-god-in-the-machine-ai-and-the-future-of-military-authority/
0 Comments
Your opinion matters, your voice makes us proud and happy. Your words are our motivation.