Pokémon Go players unwittingly contributed to tech with military drone uses
Niantic’s post-Pokémon Go pivot is turning out to be more than a corporate reshuffle. Ars reports that Niantic Spatial has been using billions of real-world images captured by millions of players, plus Scaniverse uploads, to train a “large geospatial model” that can interpret physical spaces for navigation.
That matters because the data wasn’t collected for robotics or defense; it came from a mass-market AR game and a companion scanning app. Niantic says the scans were of public points of interest and that the model is trained on the data rather than exposing the raw scans, but the end use is still broad enough to include delivery robots and possibly military drones.
The timeline is also notable. Niantic Spatial was spun out in May 2025, after Niantic sold its licensed games, including Pokémon Go, to Scopely. Before that sale, Niantic had already publicly described plans to use player scans to build a 3D model of the physical world.
For game developers, this is a useful reminder that telemetry, UGC, and location capture can outlive the original product and be repurposed in ways players may not expect. If your game collects real-world imagery or spatial data, the questions aren’t just technical anymore; they’re about consent, retention, downstream licensing, and whether your data pipeline could become part of someone else’s robotics stack.
“A large geospatial model”
- what
- Niantic Spatial is using player-captured real-world images to train a geospatial AI model for navigation.
- who
- Niantic Spatial, spun out of Niantic; data came from Pokémon Go players and Scaniverse users.
- when
- Niantic Spatial was spun out in May 2025; the Pokémon Go sale to Scopely happened before that.
- impact
- Shows how game-collected spatial data can be repurposed for robotics and possibly military applications.
Interesting tech, but troubling downstream use and consent issues.
Follow AI updates
See relevant stories in your personalized news feed.
Discussion