Since 2021, Pokémon Go has encouraged its players to capture short smartphone videos of real-world locations called PokéStops in exchange for in-game rewards, and millions of players spent years doing exactly that, documenting streets, parks, buildings, storefronts, and city landmarks from every angle, in different lighting conditions, and at different times of day. Those roughly 30 billion environmental scans are now owned by Niantic Spatial, the geospatial AI company that spun off from Niantic in May 2025 after Niantic sold its gaming titles including Pokémon Go to the Saudi-backed gaming publisher Scopely. Niantic Spatial used those scans, along with data from its Scaniverse app, to build what it calls a Large Geospatial Model, a high-fidelity 3D representation of the physical world built primarily from geolocated images taken by players who believed they were just playing a game.

The navigation system built on that data is called a Visual Positioning System, or VPS, which works by comparing what a camera sees in real time against the 3D model of the environment, and can pinpoint a location with high accuracy using as few as two recognizable reference points even when they are only a few pixels wide in the camera frame. That capability makes it particularly valuable in environments where GPS signals are jammed or unavailable, which is exactly what Niantic Spatial’s CTO Brian McClendon described as one of the system’s key advantages for robots operating in areas where signals are intentionally blocked. In December 2025, Niantic Spatial announced a formal partnership with Vantor, a defense and intelligence firm formerly known as Maxar, to integrate the VPS with Vantor’s Raptor system, a platform-agnostic, vision-based navigation software already used in autonomous drones. The combined system is designed to allow both aerial and ground platforms to navigate and coordinate precisely in GPS-denied environments, which the defense industry considers a critical capability for modern battlefield operations where GPS jamming is increasingly common.

Vantor’s defense credentials make the partnership hard to describe as purely civilian. In May 2026, Vantor was awarded a $70 million contract by the National Geospatial-Intelligence Agency, which is part of the U.S. Department of Defense, and in February 2026, just before Operation Epic Fury and the start of the 2026 Iran War, the company received a $217 million contract from the U.S. Army to support its Synthetic Training Environment program. Vantor has denied that it plans to use raw Pokémon Go player data in active deployments. However, the defense contractor conspicuously declined to state whether the core foundational navigation model it is currently fielding was trained using those 30 billion player scans in the past.

This corporate double-speak is the heart of the issue: once crowd-sourced data is baked into an AI's foundational model, the data is permanently laundered. The players' data is already inside the machine, whether the military admits to using the raw files or not.

TLDR; We are cooked ;D

https://arstechnica.com/ai/2026/06/pokemon-go-players-unwittingly-contributed-to-tech-with-military-drone-uses/

Here is a YT vid about it https://www.youtube.com/watch?v=Vk95ZO-WAvY

Posted by Available_Air_6367

3 Comments

  1. Available_Air_6367 on

    SS: For years, hundreds of millions of people thought they were just hunting virtual creatures and scanning locations for in-game rewards. In reality they were providing free, crowd-sourced labor to map the physical world down to the centimeter…basically creating a weaponized AI GeoGuessr vastly superior to Google Street View.

    While technically buried in the fine print of the TOS (as usual), players had no idea this geospatial AI model would be spun off and handed over to defense contractors to navigate military hardware in GPS-denied war zones. I haven’t seen this discussed much here, so I wanted to post 🙂

Leave A Reply