AI & Society

AI and War: How Artificial Intelligence Is Changing Warfare—and What Businesses Must Learn From It

Autonomous drones, AI targeting, and hybrid warfare 2026: what military AI deployment means for ethics, regulation, and enterprise AI.

Author: SIMO GmbH
Published on: March 5, 2026
Reading time: 11 minutes
Category: AI & Society

Artificial intelligence increasingly decides who lives and who dies. This is not a dystopian future vision but reality on the battlefields of Ukraine and the Middle East. Drones navigate autonomously through enemy jamming terrain, algorithms generate target lists with thousands of names in seconds, and governments invest billions in a technology whose ethical boundaries have not even begun to be defined. This article analyzes the current state of military AI, names the risks, and draws the connecting lines to what businesses must consider when deploying AI—because the same questions about control, transparency, and accountability arise in the civilian context as well.

The New Reality: AI on the Battlefield

Ukraine: The Testing Ground for Autonomous Weapons

Since 2022, Russia’s war of aggression against Ukraine has evolved into the largest open-air laboratory for networked, remotely controlled, and increasingly autonomous weapon systems. The numbers are impressive—and alarming in equal measure.

Ukraine is planning the production of approximately 4.5 million FPV drones for 2025—more than double the previous year. According to a study by the Royal United Services Institute (RUSI), tactical drones are responsible for roughly two-thirds of Russian losses, making them twice as effective as all other weapon systems combined.

But the real revolution lies not in quantity but in intelligence. The hit rate of conventional FPV drones is 30 to 50 percent, and for inexperienced pilots only around 10 percent. AI-controlled systems are expected to raise this rate to approximately 80 percent. By summer 2025, more than 100 operations using swarm technology had been conducted—groups of drones that autonomously coordinate attacks.

The clearest demonstration of military AI capability to date was Operation “Spinnennetz” (Spider Web) in June 2025: a Ukrainian strike on Russian military airfields thousands of kilometers inland. Remote control was impossible at this distance—instead, artificial intelligence took over navigation and the precision strikes. At least 20 Russian strategic aircraft were destroyed or damaged.

Israel’s Lavender: When an Algorithm Creates the Kill List

The most controversial example of military AI is the “Lavender” system of the Israeli Defense Forces. Developed by the elite Unit 8200, the algorithm automatically identifies human targets based on surveillance data.

The facts published by +972 Magazine in April 2024, based on statements from six Israeli intelligence officers, are alarming:

Lavender listed up to 37,000 Palestinian men as potential targets
The average human review time per target was 20 seconds—merely to confirm that the target was male
The system had an acknowledged error rate of 10 percent
For each marked junior operative, the killing of 15 to 20 civilians was deemed acceptable
For high-ranking targets, the army approved the killing of more than 100 civilians in several cases

A companion system called “Where’s Daddy?” geographically tracked the marked individuals to strike them in their family homes. Together, these systems automated the entire so-called “kill chain” from identification to attack.

UN Secretary-General Antonio Guterres expressed being “deeply troubled” by the reports. UN Special Rapporteur Ben Saul stated that many Israeli strikes in Gaza under these circumstances could constitute “war crimes through disproportionate attacks.”

Helsing: Germany’s AI Drone Champion

Germany has long been part of this development. The Munich-based company Helsing, founded in 2021, is Europe’s most valuable defense startup at a valuation of 12 billion euros. The funding rounds read like a business thriller: 209 million euros (2023), 450 million (2024), most recently 600 million euros in a Series D round led by Prima Materia of Spotify founder Daniel Ek.

The Bundeswehr has approved the purchase of kamikaze drones from both manufacturers—Helsing and Stark Defence—for initially 540 million euros, with a total potential of up to 2 billion euros. The HX-2 from Helsing, an AI-controlled strike drone with a 100-kilometer range, is already deployed in Ukraine. Its integrated AI is designed to harden the system against electronic countermeasures.

However, there are question marks: in January 2026, a hit accuracy of just 35 percent in combat was reported. And Helsing’s plans, classified as “top secret,” envision drone fleets with over 1,000 kilometers of range that autonomously conduct reconnaissance and execute mission objectives.

The Ethical Dimension: Control, Accountability, Human Dignity

The Accountability Vacuum

Autonomous weapon systems create what philosopher Robert Sparrow calls a “Responsibility Gap”: harm occurs without an identifiable moral agent who can be held accountable. Who bears the blame when an algorithm kills 15 civilians to eliminate a supposed target—the programmer, the commander, the algorithm itself?

Researchers at Queen Mary University of London concluded that the integration of AI-powered weapon systems promotes the objectification of human targets, leads to increased tolerance for collateral damage, and weakens the moral agency of operators through automation bias.

Speed Versus Judgment

A core problem: AI massively accelerates the kill chain. Where analysts once identified 50 targets per year, a system like Lavender generates 100 targets per day. This speed creates pressure that degrades human oversight to a final formality—a click in the interface, 20 seconds of review time.

Precisely this pattern—automation bias, over-reliance on machine outputs, erosion of human oversight—was described in an article in International Affairs in January 2026 as the central danger of AI-powered decision support systems. The insight applies far beyond the military.

Human Control: Demand Without Enforcement

At a meeting in February 2026 in Paris, 25 European states agreed: the “decision over life and death must not be left entirely to autonomous weapon systems.” The EU Parliament had already voted in 2018 by an overwhelming majority for an international ban on lethal autonomous weapon systems.

The reality looks different. Austria is so far the only EU member among 30 states worldwide that officially advocate a ban. Negotiations under the UN Convention on Certain Conventional Weapons (CCW) have stagnated since 2014. And the Trump administration terminated the “Political Declaration on Responsible Military Use of AI” in 2025, which had previously been signed by 45 states.

UN Secretary-General Guterres has set 2026 as the deadline for a legally binding agreement under international law. UN talks in Geneva in March and August 2026 will show whether that is realistic. The odds are poor.

The Arms Race: Numbers, Markets, Players

The financial dimensions of the military AI market are enormous:

Metric · Value
Helsing valuation (2026) · 12 billion euros
Bundeswehr drone budget · up to 2 billion euros
Ukraine drone production 2025 · 4.5 million units
Global defense AI market (forecast 2030) · over 50 billion USD
Helsing total funding · 1.4 billion euros
Stark Defence order volume · up to 2.9 billion euros

The major players are the USA, China, Israel, Russia, the United Kingdom, and increasingly the EU states. German companies like Helsing (Munich), Stark Defence (Munich), and Rheinmetall are actively positioning themselves. The Bundeswehr plans to establish six drone units with 60 to 250 soldiers each—the first should be operational by 2027, the remaining five by 2029.

Hybrid Warfare: Why Businesses Are Also Affected

Military AI is not isolated. The use of AI in hybrid warfare—cyberattacks, disinformation, supply chain sabotage—directly affects the civilian sector.

An article on netzpalaver.de from March 2026 warns: when geopolitical tensions increase, digital control layers—especially those linked to automation and decision support—become strategic terrain. Companies that operate AI systems without adequate security architecture become attack surfaces.

In concrete terms, this means:

Supply chain attacks on AI models and their training data
Adversarial attacks on decision systems
Deepfake-based social engineering attacks on executives
Manipulation of AI-powered business processes through targeted data injection

What the Military AI Debate Means for Businesses

The ethical and practical questions raised by military AI are fundamentally the same questions every company deploying AI must answer:

1. Human Control Is Not Optional

When an algorithm decides about life and death in 20 seconds, that is no longer human control. The same logic applies to business decisions: when an AI system automatically rejects credit applications, evaluates employees, or classifies customers—without a human understanding and being able to review the decision basis—that is not responsible AI deployment.

The EU AI Act requires exactly this for high-risk AI systems: human oversight, traceability, documented decision processes.

2. Error Rates Must Be Named

Lavender had an acknowledged error rate of 10 percent—and was still deployed extensively. In businesses, something similar happens: AI systems go live without error rates being systematically captured, communicated, and evaluated. The KI-MIG, Germany’s implementing law for the EU AI Act, requires exactly this transparency from 2026.

3. Governance Prevents Shadow AI

In the military, uncontrolled AI deployment leads to civilian casualties. In businesses, Shadow AI—the uncontrolled use of unauthorized AI tools—leads to data protection violations, compliance breaches, and reputational damage. The solution in both cases is the same: a binding governance framework with clear rules, responsibilities, and audit processes.

4. Transparency Builds Trust

The secrecy surrounding systems like Lavender has massively damaged trust in military AI. Companies that deploy AI must take the opposite path: transparency about deployment purposes, data sources, decision logic, and error rates is not a weakness but a competitive advantage.

Practical Checklist: Responsible AI Deployment in Business

Derived from the lessons of military AI—ten principles for enterprise deployment:

Human-in-the-Loop: No automated decisions without a human review instance for business-critical processes
Document error rates: Every AI system must transparently report its accuracy and error rate
Maintain audit trails: All AI decisions must be traceable and recorded in an audit-proof manner
Bias testing: Regular checks for systematic biases
Governance framework: Clear rules on who may use which AI tools for which purposes
Data sovereignty: Control over which data flows into which AI system
GDPR and EU AI Act: Compliance is not a project but an ongoing process
Training and AI literacy: Employees must be able to critically evaluate AI outputs
Emergency processes: What happens when an AI system produces erroneous decisions?
Regular reviews: AI systems degrade—continuous monitoring is mandatory

Frequently Asked Questions

Which countries use AI in warfare most extensively?

The USA, China, Israel, and Russia are the leading nations in military AI deployment. Israel has conducted the most extensively documented use of AI for automated target acquisition with the Lavender and Gospel systems in Gaza. Ukraine deploys AI-controlled drones at scale, and European startups like Helsing and Stark Defence supply the technology.

Is there an international ban on autonomous weapons?

No. Despite years of negotiations under the UN Convention on Certain Conventional Weapons (CCW), no legally binding agreement under international law exists. 119 states favor negotiations, but resistance from major military powers blocks progress. UN Secretary-General Guterres has set 2026 as the deadline, but experts are skeptical.

What does the EU AI Act have to do with military AI?

The EU AI Act primarily regulates civilian AI applications. Military systems are largely exempt. But the core principles—human oversight, transparency, risk assessment—are identical. For companies developing both civilian and dual-use technologies, the boundaries are increasingly blurred.

Why should businesses care about military AI?

Because the ethical questions are the same. Automation bias, lacking transparency, uncontrolled scaling, governance gaps—these problems occur in both military and civilian AI. The military examples merely show in extreme form what happens when AI is deployed without adequate control mechanisms.

How do businesses protect themselves against AI-based hybrid warfare?

Through robust security architectures: data encryption, access control, monitoring of AI systems for adversarial attacks, regular penetration tests, and employee training against deepfake-based social engineering. A governance framework is the foundation.