When Pokémon Go exploded onto smartphones in 2016, thousands of us, as players, wandered through parks, city streets and neighbourhoods hunting virtual creatures layered onto the real world. To complete tasks, players had to scan real-world locations to earn in-game rewards. It was just a game, what could really go wrong?
Those in-game scans are now, nearly a decade later, at the centre of a debate about artificial intelligence, data ownership and the increasingly blurred line between consumer technology and military applications.
According to reporting by Dutch newspaper Trouw and later detailed by The Guardian, location scans collected through Pokémon Go have helped train AI systems capable of understanding and navigating real-world environments — technology that could eventually be used by military drones operating in areas where GPS signals are unavailable.
Where Did It Begin?
The story begins with Niantic, the company behind Pokémon Go. The game became a global phenomenon after its launch in 2016, attracting hundreds of millions of players, and by 2018, Niantic said the game had been downloaded more than 800 million times worldwide.
In 2021, the company introduced a feature allowing players to scan real-world locations known as PokéStops. Users could earn rewards by recording 360-degree videos of their surroundings using their smartphone cameras. The feature was voluntary and required players to opt in.
These scans captured not only streets and public landmarks but, in some cases, private spaces as well. One Dutch player interviewed by Trouw said he had scanned parts of his apartment without considering how the footage might later be used.
Those scans became part of a vast database used to train what Niantic calls “spatial intelligence” models. According to Trouw, nearly 30 billion scans contributed to the creation of detailed AI systems capable of recognising and mapping physical spaces.
ALSO READ: Blueprint For Job Loss? How Camera-Strapped Indian Workers Are Showing AI How To Work
Where Do The Drones Come in?
The technology eventually became part of Niantic Spatial, a company spun out from Niantic after the gaming business was sold in 2025 to Saudi-owned Scopely for $3.5 billion.
Niantic Spatial’s ambition extends far beyond gaming, developing what is known as a Visual Positioning System, or VPS. Unlike GPS, which relies on satellite signals, VPS uses cameras and AI to determine location by analysing visual cues in the environment.
That capability drew the attention of Vantor, an American company specialising in navigation software for drones and autonomous systems. In December, Niantic Spatial and Vantor announced a partnership aimed at improving navigation in GPS-denied environments.
According to the companies, the technology could help autonomous systems maintain situational awareness when satellite signals are jammed, spoofed or blocked.
Modern Warfare And Ethics
Vantor has openly described GPS disruption as one of the biggest challenges facing autonomous systems. The company later secured a deal worth up to $217 million with the US Army for training software.
Both Niantic Spatial and Vantor told The Guardian that Pokémon Go scans themselves were not handed over to Vantor. However, they acknowledged that the scans were used to train Niantic Spatial’s foundational AI models.
Jeroen van den Hoven, a professor of ethics and technology at TU Delft, told Trouw that while it may be impossible to identify exactly which scans contributed to the final system, the technology likely would not have advanced as quickly without the vast amount of data generated by players.
Tom Sulston of Digital Rights Watch told The Guardian that most consumers do not fully understand how data collected for one purpose can later be repurposed for entirely different industries. Terms and conditions may technically allow such uses, but few users read lengthy legal agreements before downloading a game.
AI Economy
Companies increasingly rely on data generated through everyday activities to train sophisticated models. Experts interviewed by both publications pointed to previous examples. Fitness platform Strava inadvertently revealed the locations of military installations through publicly available activity maps. Meta’s smart glasses collect visual information from their surroundings, while self-driving car companies create detailed maps of cities using onboard cameras.
The controversy does not necessarily mean the technology itself is inherently problematic. Visual positioning systems also have civilian uses, including robotics, autonomous delivery vehicles and navigation tools. Earlier this year, Niantic Spatial announced a partnership with Coco Robotics, whose delivery robots operate in several cities.
But the episode raises larger questions about consent and transparency in the age of AI. The scans were collected legally, users opted in and the company says it operates responsibly. Yet, it’s possible that data gathered for entertainment can take on an entirely different life years later.
ALSO READ: Meta Rolls Out ‘AI Mode’ For Facebook, Answers Pull From Public Content Across Platforms
Essential Business Intelligence,
Sharp Market Insights,
Practical Personal Finance Advice, Daily Fuel, Gold and Silver Prices and Latest Stories — On NDTV Profit.
