While You Caught Pokémon, Your Data Became a Military Drone's Eyes

Remember chasing a Pikachu down the street, phone held high? Now imagine that what you actually captured wasn’t a Pokémon at all, but a precision 3D map of your neighborhood. Over the past few weeks, a connection between Niantic — the studio behind Pokémon Go — and military drone navigation technology has started circulating online. It’s the kind of thing that’s more unsettling precisely because it wore the mask of a game.

Let me be straight with you first. This hasn’t exploded across the big communities yet. Direct discussion over the last 30 days is thin, and it’s only just starting to spread through corners like Spanish-language YouTube. So this isn’t a “confirmed scandal” piece. It’s a structural argument for why this story is both plausible and genuinely dangerous.

Pokémon Go Was Never a Game. It Was a Scanner.

Here’s the core of it: Niantic’s real asset was never the game. It was the data.

Pokémon Go got hundreds of millions of players worldwide to walk through streets, parks, buildings, and alleys with their cameras up — all under the banner of augmented reality. And the whole time, those smartphone cameras were quietly photographing and scanning the space around them. Niantic calls this its VPS, or Visual Positioning System.

The simplest way to put it: GPS knows you’re “near this building.” VPS knows you’re “standing at the third-floor window of this building, facing this direction.” Stitch together the photos hundreds of millions of people upload every day, and you get a centimeter-level 3D map of the physical world — including the indoor spaces and back alleys GPS can’t reach. A map we built for free, in exchange for a little fun.

A Data Factory Called “Spatial Computing”

Niantic didn’t let all that data sit idle. The company gave it a name: the LGM, or Large Geospatial Model.

If the term feels unfamiliar, think of it this way. If ChatGPT is a large language model (LLM) trained on text, an LGM is a model trained on the physical world’s space. Instead of words, it learns roads, buildings, and terrain. And like its language-model cousins, it can infer what a place it has never seen probably looks like, based on the patterns it has learned.

This is where the story walks out the door of the game. Understanding space, inferring what sits where, locating yourself from camera footage alone — this is exactly the technology self-driving cars covet. And it’s also precisely what a military drone that has to operate without GPS is most desperate for.

When GPS Goes Dark, How Does a Drone Find Its Way?

In modern warfare, GPS is no longer something you can trust. Electronic warfare is now routine. Adversaries jam GPS signals or feed fake ones (spoofing). A drone that relies on GPS alone simply gets lost.

So militaries look for alternatives. The leading one is vision-based navigation. The drone looks at the terrain through its camera, compares it against a pre-learned map, and decides “I’m right here” on its own. When GPS dies, it finds its way by sight.

But for that to work, one thing is decisive: precise, wide-ranging ground-level visual data. Military satellites are great at the view from above. What they struggle to get is the view at human eye level — the angles seen between alleys and buildings. And that exact data is what Pokémon Go players were filling in, worldwide, for free and entirely voluntarily. The missing puzzle piece was game data all along.

Vantor, and the Path From Game Data to Weapon

The other keyword in this controversy is Vantor — a name floated in the world of satellite imagery and geospatial intelligence. Merge the data satellites collect from above with the data a game gathered from below, and the picture completes itself. The rooftop seen from the sky connects to the doorway seen from the street.

A line needs drawing here, though. The cartoonish version — “the coordinates of your Pikachu got typed straight into a missile” — is overblown. The real path is far more abstract. The value isn’t any individual’s location; it’s the 3D spatial model built from hundreds of millions of data points. When game data flows toward military use, the problem isn’t one person’s coordinates. It’s that data we believed was anonymized became, wholesale, the foundation for an entirely different purpose.

That’s the crux. We consented to a game. We never consented to becoming a component of military infrastructure. The gap between those two things is exactly today’s discomfort.

What We Should Actually Worry About

The real lesson here isn’t about Pokémon Go specifically. It’s about the pattern.

“If the service is free, you’re the product” — yeah, you’ve heard it a thousand times. What’s new is that the proposition has expanded past “taste data for ads” all the way to “a precision map of the physical world.” An activity we experienced as fun was, for someone else, a meticulously designed data-collection operation. And where that data ends up is beyond our control.

Next time you install a slick new AR game, or some clever app that wants camera permission, it’s worth a moment’s thought. Is this app trying to entertain me, or trying to survey the world through me? When your camera points somewhere, are you really the only one looking?