The B-2 Spirit is basically a ghost in the sky. It costs $2 billion a copy and looks like a flying wing from a sci-fi movie. For decades, the U.S. has banked on the idea that these bombers are invisible to anyone without world-class, integrated radar networks. Then comes along a tiny Hangzhou startup called Jingan Technology, claiming its AI-powered "Jingqi" platform just live-streamed a B-2 strike over Iran in real-time.
It sounds like the end of American air superiority. If a private company can track a stealth bomber using open-source data and some clever algorithms, then stealth is dead. Right? Read more on a connected topic: this related article.
Not so fast. While the headlines make it sound like China’s AI has cracked the code, the reality is a lot messier. It's a mix of clever data scraping, aggressive marketing, and a massive shift in how "open" our secrets actually are.
The claim that shook the Pentagon
In early 2026, during the height of the conflict in Iran, Jingan Technology posted what they called a "real-time livestream of war." They claimed their AI intercepted voice communications and reconstructed the flight paths of four B-2A Spirits as they returned from a mission. More reporting by The Verge highlights comparable views on the subject.
They didn't just say they saw them. They claimed they heard them.
The company boasted that its system pulls in hundreds of billions of data points. We're talking satellite imagery, flight trackers, weather feeds, and signals intelligence. They even released audio they claimed was a direct intercept of the bombers' radio chatter.
For a moment, it looked like a private tech firm had done what the Iranian military couldn't: put eyes on a ghost.
Why the AI victory is mostly smoke and mirrors
Here's the problem. The "intercepted" audio Jingan posted? It had already been uploaded to X (formerly Twitter) by amateur aviation enthusiasts a day earlier.
The B-2 is invisible to radar, but it isn't magic. When these planes take off or land, they often use standard radio channels to talk to air traffic control. Plane spotters—people who sit outside bases with scanners—pick this up all the time.
What Jingan likely did wasn't a technical breakthrough in radar. It was a masterpiece of data aggregation.
- They took public ADS-B flight data (which many military support planes still broadcast).
- They scraped social media for reports of "loud noises" or sightings near known bases.
- They used AI to "fill in the gaps" of a flight path based on where the plane was last seen and where it needed to go.
It's basically a very high-tech version of connect-the-dots. It's impressive, sure, but it's not the same thing as having a missile lock on a stealth target.
The real danger isn't tracking, it's targeting
Don't let the "fake" audio fool you into thinking this isn't a threat. While Jingan was chasing clout on social media, another Chinese firm called MizarVision was doing something much more dangerous.
MizarVision has been releasing high-resolution satellite imagery of U.S. bases in Israel, Qatar, and Saudi Arabia. Their AI doesn't just take pictures; it automatically tags every F-22, every fuel tank, and every Patriot missile battery it sees.
U.S. intelligence officials are worried that the Iranian Revolutionary Guard (IRGC) is using this exact data.
If you're Iran, you don't need to defeat stealth in the air. You just need an AI to tell you exactly where the "invisible" plane is parked on the ground. By automating object recognition, these Chinese firms are shortening the time it takes to plan a strike. What used to take a team of analysts hours now takes an algorithm seconds.
China’s civil-military fusion is working
You have to understand how these companies operate. They aren't just "startups" in the Silicon Valley sense. They're part of China’s "civil-military fusion" strategy.
Beijing wants its private sector to take the risks. If Jingan Technology makes a bold claim and gets caught exaggerating, the Chinese government can just shrug. But if the tech works, the People's Liberation Army (PLA) gets a brand-new surveillance tool for the price of a seed investment.
In 2024, Jingan was named a "seed unicorn" in Zhejiang Province. That's code for "the government likes what you're doing." They’re being incentivized to "pit" themselves against U.S. technology.
The end of operational secrecy
The biggest takeaway here isn't that China has a secret "anti-stealth" AI. It's that the era of hiding military movements is basically over.
We live in a world where commercial satellites from companies like Planet Labs or Maxar are constantly snapping photos. Every time a soldier posts a selfie or a "plane spotter" hits record on a scanner, data is created.
U.S. forces used to rely on the "fog of war" to stay safe. Now, that fog is being burned away by AI that can process millions of data points simultaneously. You don't need to break the physics of stealth if you can just out-calculate the logistics.
How to stay ahead in an AI-tracked world
If you're following the defense space, don't get distracted by the flashy claims of "tracking stealth." Focus on the data pipeline.
The U.S. military is already pushing commercial satellite providers to mask imagery over sensitive zones. But that's a band-aid. The real move is "algorithmic warfare" from the other side.
We’re going to see more "spoofing"—military planes broadcasting fake flight data to confuse the AI, or using electronic warfare to flood these open-source sensors with junk data.
The battle isn't just in the sky anymore. It's in the servers. If an AI can guess your flight path, you need an AI that can hide it. Expect to see the Pentagon pour billions into "counter-AI" tools that feed these Chinese startups the wrong dots to connect.
The next time you see a headline about a startup "seeing" a B-2, ask yourself: are they actually tracking the plane, or are they just really good at reading the internet? Most of the time, it's the latter. But in 2026, being good at reading the internet is a weapon all its own.
