New reports have revealed that location scans collected through Pokémon Go helped train artificial intelligence models capable of understanding and interpreting real-world environments. Those same AI capabilities are now connected to technology being developed. What’s unnerving for users whose data was collected is that they are using that data for drone navigation in military settings. Including areas where GPS signals are unavailable or intentionally disrupted.
Pokémon Go, launched in 2016, encouraged players to explore the real world in search of Pokémon. Several years later, the game introduced a feature that rewarded users for scanning real-world locations such as landmarks, parks, statues, and public spaces. Players then uploaded these scans through their smart devices, helping create an enormous database of real-world imagery and spatial information.
The Pokémon GO AR Scanning task (also known as an AR Mapping task) is a field research quest that requires you to record a short video of a real-world PokéStop or Gym using your phone’s camera. Niantic uses these scans to build detailed 3D maps and spatial AI databases.
The Data in Question
The data became part of a larger effort by Niantic, the company behind Pokémon Go, to build advanced spatial AI models. These models would be capable of recognizing and mapping physical environments. According to reports, billions of scans contributed to training these systems. Thus creating a massive digital understanding of real-world locations.
The controversy emerged after Niantic Spatial, a company formed from Niantic’s spatial computing business, partnered with Vantor. Vantor works extensively with the military, functioning as a major contractor and spatial intelligence provider for the U.S. Department of Defense. The goal of the partnership is to improve navigation in environments where GPS signals cannot be relied on. Areas where it’s impossible to navigate because of jamming, interference, or signal loss. Such conditions are common in modern conflict zones.
While both companies have stated that Pokémon Go scan data itself is not being directly shared with Vantor, they acknowledge that scans collected from players were used to train the underlying AI models that form part of Niantic Spatial’s technology stack. In other words, the raw data may not be changing hands, but the intelligence derived from that data is contributing to systems that could ultimately be deployed in military applications.
This distinction has done little to calm critics.
Many users would have never agreed to their data collections being used like this. However the scanning feature required participation and was covered by terms of service agreements. Critics contend that legal consent and informed understanding are not necessarily the same thing.
Military interest in AI-powered navigation is growing rapidly because modern battlefields are becoming increasingly harder for GPS-based systems. Jamming and signal disruption have become common tactics in conflicts around the world. AI systems capable of recognizing terrain, buildings, and landmarks can provide an alternative means of navigation. Especially when traditional positioning systems fail. These capabilities are particularly valuable for autonomous drones operating in contested environments.
A New Reason to Read those “Terms of Service” Notices
The Pokémon Go case serves as a reminder that data collected today for seemingly harmless purposes may have consequences. Ones that extend far beyond its original intent. For millions of players who scanned landmarks in exchange for in-game rewards, the possibility that their contributions helped train technology with military applications raises questions about transparency, consent, and the future uses of personal data.
The larger issue is whether consumers truly understand how the data they generate can be transformed into technologies that eventually shape the battlefield. It’s uncomfortable to think that your data could have been used for hurtful purposes. The information gathered may be for dropping supplies or dropping bombs. However users would never be able to know what exactly the data was used for. It’s kind of a crisis of consciences for everyone involved.
