AI Turns War into a Game: U.S. Military Strikes 1,000 Targets in Iran in 24 Hours, "Machine Speed" Becomes the Standard

U.S. military strikes over 1,000 targets within the first 24 hours of operations against Iran, nearly twice the scale of the 2003 Iraq War. Powered by deep-integrated AI targeting systems like Maven Smart System, one of the language models used is Anthropic’s Claude.
(Background: The Wall Street Journal reports: Trump used Claude AI for Iran Hamedani targeting, OpenAI fully takes over Pentagon systems)
(Additional context: Pentagon confronts Anthropic! Fully opens Claude for military use—“or else terminate contract”)

Table of Contents

Toggle

• This is not human speed
• Maven Smart System: The battlefield brain
• Decision steps compressed to the extreme
• Banned, then used to strike Iran
• From “faster possible” to “must be faster”

Using AI to assist firepower in U.S. strikes on Iran is not surprising in itself.

What is astonishing is the speed.

On February 27, 2026, the U.S. and Israel launched a joint operation codenamed “Epic Fury” against Iran. Though the name sounds dramatic, within the first 24 hours, over 1,000 targets were struck.

The entire operation extended to over 2,000 targets. U.S. Central Command General Brad Cooper described the scale as nearly double that of the 2003 Iraq invasion.

Doubling isn’t about quantity—it’s about time.

This is not human speed

During the 2003 Iraq War, it often took tens of hours for a target to go from intelligence confirmation to strike approval. Satellite imagery was reviewed frame by frame manually, intelligence reports were passed between departments, translated, reported, and re-verified. This is “human speed,” limited by attention span, working hours, and cognitive capacity (memory).

But now, it’s different.

Maven Smart System integrates data from over 150 sources, including satellite imagery, drone footage, signals intelligence (SIGINT), human intelligence (HUMINT), and historical strike records. These data streams are input into the system every millisecond. Claude is one of the core language models that helps Maven interpret this data.

The specific workflow is: AI handles initial filtering and semantic understanding, completes preliminary analysis, assigns target priority levels, and pushes this to human analysts. Analysts no longer start from scratch drowning in data. They only need to review AI-predicted, pre-ranked information, and make the final decision.

U.S. Central Command spokesperson Captain Timothy Hawkins said: AI technology “can perform initial filtering of incoming data, allowing analysts to focus on higher-level analysis and verification.”

Humans decide on strategic choices, AI provides the rapid data processing.

Maven Smart System: The battlefield brain

Maven Smart System isn’t new. Developed by Palantir Technologies, it originated from the 2017 U.S. Department of Defense “Project Maven,” which sparked massive internal protests at Google and eventually led Google to withdraw from military AI contracts.

But Palantir persisted.

Today, Maven has become one of the most deeply integrated AI platforms within U.S. military networks. Claude can operate on Pentagon’s classified networks—this is a result of the 2024 contract between Anthropic and the Department of Defense, capped at around $200 million for two years. Claude is thus the first commercial AI model permitted inside U.S. military classified networks.

In the Iran operation, Maven’s specific functions included:

• Real-time aggregation of multi-source intelligence data (satellites, drones, signals interception)
• Automatic ranking based on target importance and threat level
• Generating precise coordinates and strike recommendation lists
• Allowing human analysts to focus on verification and final authorization, not data sorting

If traditional intelligence systems are a library, Maven is an assistant that has read all the books, summarized them, and sorted them by importance. You just tell it what you need, and it already knows where to find and what to deliver.

Decision steps compressed to the extreme

Traditional military decision chains roughly follow:

Sensors collect data → Analysts read and organize → Submit reports → Commanders evaluate → Legal review → Authorization for strike

Each step involves waiting—waiting for translation, review, reporting, confirmation. This process retains significant “decision buffer time,” but also means the pace is limited by the slowest link.

Maven + Claude compresses the “sensor data collection → analyst review” process from human time to machine time. AI doesn’t replace final strike decisions but significantly shortens the time to reach that decision point.

In the first 24 hours, 1,000 targets were hit—about 41 targets per hour, less than one per minute. At first glance, not very fast, but behind each target’s intelligence gathering, location confirmation, threat assessment—these tasks, under AI assistance, are compressed into an efficiency that humans alone cannot achieve.

The speed of war has been fundamentally altered by AI.

Banned, then used to strike Iran

Here’s one of the most peculiar tech news stories of 2026.

On February 27, 2026, the Trump administration announced a ban on Anthropic, citing “supply chain risks” and “national security threats,” ordering all federal agencies to immediately cease using Claude. The reason: Anthropic refused to remove safety guardrails from Claude—especially those that prevent its use in autonomous weapons control and mass surveillance.

Further reading: Anthropic CEO blasts: OpenAI and Pentagon contracts are all lies, Altman pretends to be a peace envoy

Almost simultaneously, the U.S. military was using Claude to analyze targets for the “Epic Fury” operation.

Banned. Yet still used to strike people—on the same day.

The Pentagon later explained there was a “six-month transition period.” But anyone familiar with operational systems knows you can’t just swap out core AI systems mid-mission. Military commanders are deeply dependent on Maven + Claude workflows; replacing them isn’t just a technical issue but a risk management concern.

This is the reality of AI deeply embedded in military systems: policies can ban a company, but operationally, removal isn’t quick or easy.

From “faster possible” to “must be faster”

One thing is certain: the speed advantage brought by militarized AI won’t disappear just because a company loses a military contract.

If Claude is removed, Maven will integrate OpenAI’s models. In fact, after Anthropic’s ban, OpenAI immediately announced it would take over the Pentagon’s AI systems. The logic of speed remains unchanged—only the provider changes. Come on, Sam Altman is eager to grab that seat.

This is the core significance of understanding the recent Iran operation.

Human warfare has never been static. Gunpowder changed the meaning of walls, aircraft altered the concept of frontlines, nuclear weapons redefined the bottom line of war. Every technological leap brings both “efficiency gains” and “amplified errors.” The faster the pace, the less time there is to recover.

AI is changing the length of the decision chain. When a system can assist in integrating and prioritizing 1,000 targets’ intelligence within 24 hours, war no longer waits for the slowest human link. Moreover, the U.S. military possesses untested (or at least unadmitted) fully autonomous weapon systems capable of identifying, deciding, and executing strikes independently.

This Iran operation may be the first public example of AI-assisted warfare. “Machine-speed warfare” is already happening, and there’s no turning back.