The people behind Pokémon Go: Conducting global mapping for military AI.

Author: Sleepy.txtThere have been several large-scale geographical data collection operations in human history, each of which has changed the power structure.During the Age of Exploration, Portugal and Spain sent fleets to draw nautical charts. Whoever controlled the accurate sea routes controlled the initiative in trade and colonization.During World War II, the U.S. Army Mapping Office produced military maps covering the entire globe. Eisenhower later said that maps were a key factor in the Allied victory.During the Cold War, the U.S. and the Soviet Union used spy satellites to photograph each other's territories. Image analysts spent their days staring at blurry satellite photos, counting the number of enemy missile silos. These three operations share a common thread: they were all state-sponsored, conducted in secrecy, and required enormous military budgets, all unknown to the general public. In the summer of 2016, 500 million people voluntarily picked up their mobile phones, went into parks, streets, and shopping malls, and used cameras to scan their surroundings, playing a game called Pokémon Go. Unbeknownst to them, they were participating in the largest, lowest-cost, and most efficient geographic data collection operation in human history. Every time they went out, they were uploading ground-view images of buildings around the world to servers free of charge. John Hanke, the creator of Pokémon Go, initially intended it simply to be a fun AR game. However, ten years later, he made a different choice, directing this data towards AI and military applications. Behind this choice lies a chilling clue: the first funding Hanke received when he started his business in 2001 came from the CIA. The CIA's First Funding and a Young Man's Map DreamIn 1996, John Hanke received his MBA from the Haas School of Business at UC Berkeley, and then did something that seemed quite reckless for the time: start an internet business.His first company was called Archetype Interactive, which developed a role-playing online game called Meridian 59, one of the earliest 3D online games in history, almost ten years before World of Warcraft. This company was later acquired by 3DO, and Hanke earned his first fortune.In 2001, he founded Keyhole. He had a vision for this company: to stitch together satellite images into an interactive globe, allowing anyone to see any corner of the Earth on a computer screen. At that time, the average internet speed in American households was 56Kbps, meaning loading a single high-resolution image took half a minute. Keyhole's product required real-time streaming of satellite images, making the technical challenge immense. However, one organization immediately recognized its value: In-Q-Tel, the CIA's venture capital arm. Founded in 1999, In-Q-Tel was personally spearheaded by CIA Director George Tenet. Its operating logic was never about making money, but rather about enabling intelligence agencies to introduce cutting-edge civilian technologies into their systems through commercial investment. At that time, satellite imagery only showed the top of buildings, while intelligence analysis required a ground-based perspective, such as the sides of buildings, street layouts, entrance and exit locations, and the obstruction of the surrounding environment. Keyhole's work perfectly met this need. In 2003, weeks before the outbreak of the Iraq War, In-Q-Tel completed its investment in Keyhole. Soon, Keyhole's technology was put into practical use for battlefield situation analysis; military analysts used it to study the street layout of Baghdad and the architectural structure of Saddam Hussein's palace. In October 2004, Google acquired Keyhole for tens of millions of dollars. Keyhole's core technology became Google Earth, which later changed the world, and Hanke himself joined Google as Vice President of Geographic Products, leading a larger-scale project. Around the same time, Peter Thiel used the tens of millions of dollars he earned from selling PayPal to found a company called Palantir. The company name comes from the crystal ball in Tolkien's *The Lord of the Rings*. Palantir's idea was to use big data analytics to help intelligence agencies find traces of terrorists within massive amounts of information. While mainstream Silicon Valley venture capitalists scoffed at it, In-Q-Tel became Palantir's earliest endorser and investor, and the CIA itself became Palantir's first client. The "Mistake" of the Street View Car and a Culture of Collecting First and Apologizing Later After joining Google, Hanke led the Street View project, which later sparked global controversy. This was the first time in human history that cameras were used to record streets around the world. Google designed a special car with a 360-degree camera array on its roof, which began driving around the world in 2007, covering more than 60 countries and taking billions of photos. These photos were stitched together to create continuous street views, allowing anyone to walk onto any street on Google Maps. This was a product that revolutionized the way humans perceive space. But in 2010, a scandal completely changed public perception of the project. The Hamburg Data Protection Authority, a German regulatory body, discovered during an inspection that these Street View cars were secretly intercepting Wi-Fi data packets from homes and businesses along their routes while simultaneously photographing the streets. The intercepted data streams included passwords, email content, medical records, banking information, and even the content of web pages being viewed. This practice violated communications privacy laws in Germany, France, Australia, the UK, Canada, Spain, and many other countries. Google faced massive fines globally, the US Federal Trade Commission launched an investigation, and the US Department of Justice also became involved. Ultimately, Google paid a $7 million settlement in the United States, was fined €100,000 in France, and also paid fines in Germany, Australia, and other countries. Faced with overwhelming evidence, Google's official explanation was a "technical error." An engineer named Marius Milner "unintentionally" introduced data interception functionality into the code, without the knowledge of management. This code ran on hundreds of vehicles worldwide for three years, collecting data from more than 30 countries, without any internal whistleblowing or management intervention. The FCC investigation later revealed that Milner had explicitly described the code's functionality in internal documents, which were distributed to multiple Google engineers. Tear off this fig leaf, and you'll see the true Silicon Valley consensus: the bottom line for data has never been a physical barrier, but rather a piece of clay that can be secretly and arbitrarily molded. The price of being caught is an apology, a fine that's negligible for Google, and then it continues.More noteworthy is Google's handling of this incident.Even more noteworthy is Google's handling of this matter. It refused to submit complete data to regulatory agencies in various countries, blocked direct investigations into Milner on the grounds of "engineer privacy," and retained the data for years after investigations in multiple countries concluded. Harboring this rogue logic of first overstepping boundaries and then paying for forgiveness, Hanke considered leaving Google around 2010 to start his own business. Pikachu was the bait; the data was the prey. In 2015, Hanke officially founded Niantic. The company initially began incubation within Google in 2010, later becoming independent and securing $30 million in funding from Google, Nintendo, and The Pokémon Company. On July 6, 2016, Pokémon Go was launched. Within a week, it boasted over 21 million daily active users, surpassing Twitter's daily active users. Within 60 days, it had been downloaded over 500 million times. In the United States, the game's daily active users at one point even exceeded those of Google Maps. People ran wildly through parks like they were under a spell, stood like stakes at intersections, and pointed their phone cameras at every corner of their surroundings, all in an attempt to catch the Pikachu on the screen. This frenzy even escalated into absurd farces: some people climbed into others' backyards to catch Pokémon, some crashed their cars into trees while staring at the screen, and one reckless individual even went to the entrance of a New Zealand police station and earnestly asked the officers if they could go in to catch a Pokémon. But if you break down each core mechanism of the game, you'll find that every design element of this mechanism maximizes the collection of user sensor data.However, if you break down each core mechanism of the game, you'll find that every design element of this mechanism is designed to maximize the collection of user sensor data.In-game missions drive players to specific locations, using cameras to scan buildings; Elf Nests allow players to repeatedly visit the same location, helping the system build multi-angle 3D models of that location; Gym mechanics encourage players to stay for extended periods, collecting detailed data on the indoor and surrounding environment; AR mode directly utilizes the camera, overlaying virtual IPs onto the real environment while simultaneously capturing images of the real-world scene. Each capture uploads more than just a photo; it's a complete sensor data packet containing GPS coordinates, phone orientation, elevation angle, and movement speed. This data is uploaded in real-time to Niantic's servers, where it's processed and gradually pieced together into a detailed global geographic database. Niantic calls this system the "Real-World Platform." Undeniably, it has brought an unprecedented AR entertainment experience to hundreds of millions of players. However, at its core, this globally popular game objectively served as the most efficient geographic data collection network in human history. Players were chasing childhood dreams, while the system quietly harvested the spatial coordinates indispensable for building a digital twin world. In its early stages, the iOS version even quietly requested full access to a Google account, theoretically allowing it to read all of a user's emails, Google Drive files, and search history. This access was publicly disclosed by security researcher Adam Reeve, drawing the attention of the US Congress. Senator Al Franken sent a formal inquiry to Niantic, demanding an explanation for the need for such extensive permissions. Niantic subsequently revoked these permissions, citing the familiar "technical error" as the explanation. At the time, governments around the world reacted remarkably quickly, with many countries banning the game outright. These countries' assessments were ridiculed by some media outlets as conspiracy theories. But in hindsight, they simply realized earlier than others that a game that had 500 million people voluntarily raising their cameras to scan buildings around the world was essentially the largest geographic data collection machine in human history. Moreover, the operating costs of this machine were entirely borne by the users themselves, including data fees, electricity bills, phone depreciation, and the time and effort spent walking. Strip away the game's outer shell, leave behind the data goldmine, and then sell it to the battlefield. Fast forward to 2025, Hanke sold Pokémon Go. The buyer was the game company Scopely, and the real investor behind Scopely was the Saudi sovereign wealth fund PIF. PIF, the Saudi sovereign wealth fund led by Crown Prince Mohammed bin Salman, manages over $700 billion in assets, making it one of the world's largest sovereign wealth funds. Its portfolio includes stakes in Uber, Lucid Motors, Nintendo, and a host of Western technology companies. This deal means that the ownership chain of historical location data accumulated by hundreds of millions of players worldwide over nine years now points to Riyadh. Why would a Saudi sovereign wealth fund spend $3.85 billion to buy a mobile game whose daily active users have declined significantly from their peak? The answer may not lie in the game itself, but in the global location data network behind it. Pokémon Go's player data contains the detailed movement trajectories of hundreds of millions of people over the past nine years—where they live, where they work, where they pass through daily, and how long they spend in which buildings. This is a dataset that any advertising company, intelligence agency, or urban planning department would dream of. But the most ingenious part of this deal is what Niantic didn't sell. Hanke retained the 30 billion images, the geodatabase that had been continuously optimized over nine years, and the large geospatial model trained on this data. He used these assets to establish a new company, Niantic Spatial, and in March 2026 publicly announced a partnership with food delivery robot company Coco Robotics. Coco's robots use Niantic Spatial's LGM (Low-Glass Mediator) for centimeter-level visual positioning when delivering food on the streets of Los Angeles and Dallas, enabling accurate navigation even in urban canyons with unstable GPS signals. However, commercial robots are merely a civilian demonstration of this technology. In December 2025, Niantic Spatial signed a cooperation agreement with defense company Vantor to use LGM for navigation of military drones in GPS-denied environments. The so-called "GPS denial environment" is an increasingly common situation on the modern battlefield, where GPS signals are jammed or spoofed through electronic warfare, rendering satellite-navigation-dependent weapon systems ineffective. In the Ukrainian theater, Russian GPS jamming equipment has caused numerous drones to lose their way. Vantor's solution involves allowing drones to "see" the ground using cameras, comparing their location with ground visual features stored in LGM. These ground visual features stored in LGM are derived from the 30 billion images taken by Pokémon Go players. A complete chain has finally surfaced. A mapping company invested in by the CIA has become the world's largest ground visual data acquisition game; this data is used to train AI geospatial models, ultimately becoming the navigation system for military drones. Around the same time, Palantir, also incubated by the CIA, had its CEO, Alex Karp, openly declaring that AI was reshaping the form of modern warfare. Palantir's AI platform, AIP, had been deployed in the Ukrainian theater, assisting the Ukrainian army in target identification and strike decisions. In Gaza, Palantir's system was used by the Israel Defense Forces for intelligence analysis. Two seeds planted simultaneously in Silicon Valley by intelligence capital—Hanke and Thiel—one collecting data from the physical world, the other analyzing data from the digital world—have converged twenty years later at the endpoint of the militarization of AI. Nobody told you this. In 2025, Hanke published an article on Niantic Spatial's official blog titled "The Next $10 Trillion in AI Investment Should Flow to the Physical World." His logic was that large language models had already taught AI to speak, but for AI to truly enter the real world—for example, to drive cars, control robots, and move autonomously in cities—it needed something else: spatial awareness of the physical world. Niantic Spatial aimed to become the Google Maps of the AI era, a shared physical world coordinate system for all robots and AI agents. This vision sounded perfectly legitimate. But it obscures a premise that has never been publicly clarified: the data for this coordinate system comes from 500 million ordinary users who never consented to the use of their location data for military applications. In 2016, the privacy policy you clicked to agree to when registering for Pokémon Go stated, "We may share data with third-party partners to improve the service." No one knew who these partners were, no one knew the boundaries of "sharing," and no one anticipated that "improving the service" would ultimately include helping drones navigate on the battlefield. When this data flowed into military applications, no regulatory body intervened, no user was notified, and no country's legal framework could trace this data transfer chain. Here's a comparison worth considering. The US government, citing concerns that "data could be accessed by the Chinese government," demanded that ByteDance forcibly divest its US TikTok operations, holding hours-long hearings in Congress. Meanwhile, the US government remained silent on the sale of Pokémon Go data to a Saudi sovereign wealth fund and its use in military applications. The logic behind this double standard has never been data security, but rather geopolitics. Data security depends on who controls it. A deeper issue is that this isn't just a Pokémon Go problem. Meta's smart glasses are constantly scanning the user's surroundings, Apple's Vision Pro is building indoor 3D maps, and Waymo's self-driving cars are reconstructing detailed models of city roads. These data collection practices are not fundamentally different from Pokémon Go's. In the AI era, every consumer device with a camera is a potential geographic data collection node, and users are almost completely unaware of this. Now, it's time to thoroughly penetrate this hidden commercial dark web. In the summer of 2016, an ordinary person, trying to catch a Pikachu, inadvertently scanned the wall of a building with their phone. This casually taken photo was silently uploaded to a server, where, along with 30 billion other similar photos, it trained a large geospatial model. This model was then housed in a new shell called Niantic Spatial—and let's not forget, this company's owner received initial funding from the CIA. Today, it acts as an electronic eye for killer drones that have lost GPS signals on the battlefield. The game itself was already priced at $3.85 billion and sold to the Saudi sovereign wealth fund, which is keen on transnational surveillance. Individually, each link in this chain of interests is impeccably clean, entirely within the bounds of compliance. But once tightly interlocked, it becomes a Leviathan that no law can restrain. This is perhaps the most thought-provoking paradox of the digital age. A system operating in the sunlight, conforming to the logic of business compliance, initially driven by people's pursuit of pure pleasure, may ultimately become part of a cold war machine under the influence of capital and geopolitics. Perhaps Hanke initially only wanted to make a good game, but under the watchful eye of intelligence capital, this data inevitably flowed to the most lucrative and bloodiest outlets. Hanke wasn't lying; he simply swallowed the bloodiest part of his statement. And those hidden lines are now crossing the horizon of some battlefield, transforming into the whistling sound of guided bombs tearing through the air. At 10:45 AM on February 28, 2026, the whistling sound appeared in Minabu, Hormozgan Province, Iran. A U.S. Tomahawk cruise missile precisely struck the Shajareh Tayyebeh Girls' Primary School, instantly collapsing the two-story building and killing 175 people, the vast majority of whom were girls aged 7 to 12. The school, located next to an Iranian Revolutionary Guard naval base, was formerly military property of the Revolutionary Guard and later converted into a school. However, in the eyes of the Pentagon's target recognition system, powered by Palantir and top-tier AI, it remained a military stronghold that had to be eliminated. No one knows whether any of the fatal architectural features in this vast database that determined the fate of nearly 200 lives were collected free of charge ten years ago by an ordinary person using a camera while sweeping the streets. But what we know is that in this meat grinder welded shut by countless cameras, Silicon Valley data brokers, and the military-industrial complex, the elixir of "technology for good" has long since lost its effectiveness. Ultimately, it only spews cold, blind, and precision-guided death upon the real world.