
AI Leak Exposes Game's Secret Character, Raising Data Privacy Concerns

By Amy Hennig


By Amy Hennig
Independent video game creators frequently encounter various challenges, such as unexpected mass refunds for their titles or the simultaneous launch of identically named games by different developers. Now, a new and unsettling obstacle has emerged: artificial intelligence sifting through private documents and prematurely disclosing unannounced game details.
This exact scenario unfolded for Klub Kofta Studio, a lone developer behind the tower defense game 'Operation Octo,' which debuted on Steam in September 2025. The developer shared on Reddit how Google's AI divulged the name of an upcoming character, a detail Klub Kofta Studio had kept entirely private. The developer explicitly stated, "I had never told the name to anyone."
Polygon reached out to both Klub Kofta Studio and Google for comments regarding this incident but received no immediate response prior to the publication of this report.
Klub Kofta Studio elaborated on Reddit that a member of their Discord community was playfully quizzing Google's AI about the game, expecting amusing responses. However, some of the AI's answers included surprisingly obscure information that, according to the developer, should not have been accessible for a small independent game like theirs.
Encouraged by these unexpected disclosures, Klub Kofta Studio requested the player to probe the AI about future, unannounced content. Astonishingly, the AI disclosed the highly specific character name 'Vantage Tripod,' a name the developer had never publicly mentioned. They noted, "As far as I know, the only place where the info exists in a digital form is inside one of my own Google Docs, and this was NOT me speaking to the AI!" Screenshots of their Discord conversation captured their bewilderment at the AI's accurate revelations.
The developer reflected, "My game is not big enough for there to be much noise out there to confuse Google's AI, so I guess that's why it managed to give actual, scarily real leaks on my game (without being confused by online speculations), and I have no idea how it knew all this." This incident underscores the AI's disturbing ability to extract precise information even from obscure sources.
Google's AI overview is known for sourcing information from platforms like Reddit, sometimes leading to inaccurate or manipulated results. For instance, the AI overview can be influenced, potentially disseminating unreliable information. Last year, a YouTuber successfully tricked Google's AI search into believing that 'GTA 6' would feature a "twerk button" through a simple online prank.
As noted by users on Reddit and Bluesky, Google automatically enrolls users into sharing their Google Drive data for purposes such as enhancing AI models. Users can adjust their privacy settings through Google support to manage how their activity interacts with features like Google's Gemini. Over recent years, major technology firms have been subtly updating their terms of service, often opting users into allowing their data to be used for AI model training.
A Reddit user voiced their concern, stating, "It's what companies do. Quietly opt all your stuff into AI while trying to avoid announcing it, then bury an option to opt out. The fuckers." Others expressed apprehension about the potential for AI overviews to leak more personal information from what users consider "private" Google Drives. One commentator mused, "Doesn't that mean you can dox people by asking very specific questions given that people's private personal data is included in the training sets then?......"
It is increasingly crucial to investigate who has access to your public and private data and how it is being utilized. The idea of retreating to a secluded, off-the-grid existence has never seemed more appealing in light of these pervasive data privacy concerns.
About the author

Renowned video game writer and director, known for her narrative work on the "Uncharted" series.

by Reiner Knizia
by Reiner Knizia