Google Gemini Allegedly Leaked Unreleased Game Content From Private Google Doc

Game Developer Claims Gemini Served Up Data Only Available in Private Google Doc

A game developer claims Google's Gemini chatbot has revealed a key detail about their upcoming game update that had only been shared in a private Google Doc, Polygon reports. Klub Kofta Studio's Operation Octo is a cutesy tower defense game about undersea life with bullet-hell elements. It's done quite well, and the solo developer was planning new content, including a new character called "Vantage Tripod." They didn't tell anyone; the name was only present in a Google Doc. But somehow, a fan chatting to Gemini heard about it. The fan was posting some of the funny responses it got from Gemini in the Discord chat with the developer, who prompted them to ask about any "upcoming volcanic ridge turrents / skeletons." Gemini responded with Vantage Tripod and a Skeleton Dragon. Both characters are upcoming, but only the Skeleton Dragon had been remotely hinted at on the game's store page. The developer insisted they didn't mention Vantage Tripod anywhere. Other responses also contained "surprisingly obscure info," the developer claimed, leading them to believe that Google had scraped his private document to train Gemini. They maintain that the game isn't big enough to have enough online discussion for the AI to interpret it in this particular way. Google says it doesn't use private Workspace content, including Drive and Docs, to train its "foundation AI models," including Gemini. In a statement provided to Polygon, it noted that "Links to publicly shared docs may be indexed if someone posted them somewhere publicly, where they could be seen by search engine crawlers." The developer admits that, in a game of Gartic Phone, they drew some elements from the game, and fans playing along called it the "Tripod Fish," which is related to the "Vantage Tripod" character they were looking to introduce. But they never mentioned that particular name. In further screenshots within a Reddit thread, the developer showed a few other examples of the AI being surprisingly accurate in describing future updates and changes that have only been saved in Google Docs. Reddit users responding to the developer highlighted how, for some people, Gemini is automatically connected to their workspace, including Docs and Drive. You can confirm whether yours is toggled to On (and change it if you wish) by navigating to the Gemini page, then Settings > Connected apps, and toggling off any integration you don't want there.

Google's AI somehow knew a game secret that existed only in a private Google Doc

Serving tech enthusiasts for over 25 years. TechSpot means tech analysis and advice you can trust. What we know so far: After a player asked Google's AI about unreleased content in his game, an indie developer says it returned a character name that existed only in a private Google Doc. Klub Kofta Studio, the solo developer behind the tower defense game Operation Octo, said the character, called Vantage Tripod, had not been announced or shared with anyone. The issue began in the game's Discord server, where a player was asking Google's AI questions about Operation Octo. The answers sometimes included details the developer considered unusually specific for a small game. The developer then asked the player to try questions about content that had not been revealed. Google's AI returned the name Vantage Tripod, according to a Reddit post by Klub Kofta Studio. "Somehow, the AI spitted out the exact & highly specific character name 'Vantage Tripod', which I had never mentioned to anyone," the developer wrote. "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!" Posts from the indiedev community on Reddit Operation Octo was released on Steam in September 2025. The developer said the game's limited online presence made the answer more alarming, not less. "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," the developer wrote. The developer later told Polygon that the character name was not included in the game's code, files, or Steam data. Google said it does not use private Workspace material, including Drive files and Docs, to train its foundational AI models, including Gemini. The company said publicly shared documents can be indexed if links appear online where search crawlers can find them. It also pointed to Google Drive's sharing controls, which let users decide who can access individual files. The available evidence does not establish how Google's AI produced the character name. It could not be independently recreated after the Reddit post circulated. The developer said that later searches began returning the name while citing the Reddit post as the source. "I'd like to say that my specific case is no longer replicable, and the AI would simply give the name 'Vantage Tripod' then point to the Reddit post as source now," the developer said. That makes it harder to separate the original response from what the AI can now retrieve from the public web. Once a claim appears in a widely shared post, AI search products can cite that post back to users, even if the original answer remains unexplained. Google's AI Overviews have increasingly used Reddit material in search results. That can be useful when users are looking for firsthand discussion, but it can also create problems when posts contain rumors, jokes, or unsupported claims. AI-generated search answers can also be manipulated. A YouTuber last year persuaded Google's AI search to repeat a false claim that Grand Theft Auto 6 would have a "twerk button." For developers, the case is a reminder that unreleased project information can be difficult to contain once it reaches a public AI answer or online discussion. It also raises a basic question for AI search systems: when a response includes obscure information, users need a clear way to know whether it came from a reliable source, a public document, or the model itself.

Google's AI Search seemingly leaked unreleased game content and no one knows how

A Google representative shared a statement denying that the company scans private Workspace content to train its AI models. There have been plenty of times when Google's AI Search has been criticized for providing inaccurate information. In some instances, those screw-ups have been highly public, like when AI Overviews recommended adding glue to pizza. However, it appears that it can also sometimes veer into being unexpectedly accurate. For example, revealing an unannounced character for a game. The solo developer behind Klub Kofta Studio alleges that Google's AI Search told a player about an unreleased character for their tower defense game Operation Octo. According to the Reddit post, the player was asking the AI "silly" questions about the game to get a laugh when it suddenly surfaced "surprisingly obscure info that should not be available for a small indie game like mine." Out of curiosity, the developer asked the player if they could ask the AI about future content for the game, and the AI returned the highly specific character name "Vantage Tripod." The developer claims they never told anyone about this character publicly. They add that the character's name is not present in any of the game files and that "As far as I know," the information only exists digitally in a private Google Docs file. "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," the developer said. It's unclear how exactly Google's AI managed to get that unreleased information. And now that the Reddit post has gone viral, any questions you ask the AI about this topic now pulls from this thread.

Is Google Gemini trained on Google docs? One indie developer thinks so, after it told players about his unreleased game plans

* A game developer claims Gemini leaked information in his private Google Docs * Per screenshots shared online, the AI told players info the dev claims was never made public * TechRadar has tried recreating the results, without success Is Google Gemini secretly scraping our Google Docs to train its AI? That's something one game developer is wondering after a player was able to learn unreleased information about their game from the AI -- including specific details the dev claims were never shared outside of a private Google Doc. According to the developer's Reddit post, specifically Google's AI knew the name of a character 'Vantage Tripod' before it had ever been released publicly -- the most the fanbase knew was that there's a 'Tripod Fish' character planned for the game, but not this exact name. It also seemed to be able to regurgitate accurate information related to unreleased mechanics that the developer says they had only written into GDocs the day before the player's AI interaction. The developer admitted Google's AI made a few mistakes, but said that some details were scarily accurate in many ways -- accurate enough that the developer is certain Google must have scrapped his private documents. Google says Gemini can access Google Docs information, but only does so when given express permission -- such as being asked to summarise a document -- and it adds that even when it does go into your files, Gemini handles data transiently. That is, Google's AI won't retain anything. There are a couple of exceptions to this. If a Google Doc is accessible to 'Anyone with the link' and that link is posted publicly online (such as on a page or in a forum) Google could scrape it for training data. Alternatively, if you've allowed a third-party extension to scrape your Google Docs, it's possible Gemini could get access to that data -- indirectly seeing what's in your docs. If you were to type out info from your docs into Google Gemini, it would also learn about what you had written that way. Despite Google's promises, some aren't entirely convinced. The thread I shared above (along with plenty of anti-AI and anti-Google subreddits) is full of people certain that Google has drained their digital files for all the data it can find. Privacy company Proton has also published an article outlining details such as Google's privacy hub not explicitly saying it won't use your content for AI training. So many unknowns I've reached out to Google with a request for comment about what has happened here (it has yet to get back to me), and while waiting for an answer I had an attempt at recreating the responses by prompting Gemini myself with no luck. It's only reference to these details was the Reddit post information. Later screenshots shared by the developer show the friend who got the AI to divulge the details originally also failed to recreate the "fluke." The whole situation is very weird and it's impossible to tell exactly what has happened, based on the information available. While it certainly looks like Gemini has used information it shouldn't have, it's equally possible that the developer has provided the AI or the friend with access to that information without realising -- allowing Gemini to respond the way it did. The AI could also have hallucinated the information in some way, or perhaps taken inspiration from previous prompts to inform its comments. Further, it's worth noting that the only responses that would stand out are these two that get very close to the truth, the sea of incorrect guesses that the AI could have given wouldn't bat an eyelid. Regardless of what happened it's another situation that reminds us to be careful with our personal info. Digitally storing personal data has its risks beyond possible AI scraping, and sharing data with an AI often means your private info won't be private anymore. Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.

Steam Developer Alleges Google Leaked Confidential Game Content via AI Search

Video game developers of any size have plenty of challenges to face, like mass refunds of their short game or two developers releasing games with the same name on the same day. A new obstacle to worry about? AI scraping private documents and leaking unannounced information about their games. That's what happened to Klub Kofta Studio, a self-described solo developer of the tower defense game Operation Octo, which launched on Steam in September 2025. In a post on Reddit, they shared Google's AI told a player the name of an upcoming Operation Octo character that Klub Kofta Studio had never publicly shared. "I had never told the name to anyone," they wrote in the post title. Polygon has reached out to Klub Kofta Studio and Google for comment, but did not hear back in time for publication. As Klub Kofta Studio explained it on Reddit, a player in their Discord server was asking the Google search AI "silly questions" about the game to get a kick out of the answers. "But the answers sometimes contained surprisingly obscure info that should not be available for a small indie game like mine." Klub Kofta Studio asked the player to throw some questions at the AI about future, unannounced content. "Somehow, the AI spitted out the exact & highly specific character name 'Vantage Tripod', which I had never mentioned to anyone," they wrote. "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!" They shared screenshots of a snippet of their conversation on Discord, including their befuddled reactions to the AI's eerily correct answers. "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," they said. Google's AI overview has a habit of pulling information from Reddit threads for answers, correct or otherwise. The AI overview can also be manipulated, which in turn can cause it to dole out unreliable misinformation. Last year, a YouTuber managed to convince Google's AI search that GTA 6 would include a "twerk button" through a fairly low-effort shitposting project. As people on Reddit and Bluesky have pointed out, Google automatically opts users into sharing their Google Drive data with Google to improve AI models, among other uses. (You can head to Google support to learn how to manage how your activity interacts with Google's Gemini features.) For years now, large tech companies have been quietly updating their terms of service and opting users into allowing the use of their data for AI model training. "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," one person commented on the Reddit thread. Others are worried about more personal information being leaked by AI overviews based on information in users' "private" (heavy air quotes here) Google Drives. "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?......" another commentor mused. If you haven't yet, now would be a good time to look into who has access to your public and private data and what they're using it for. Living in an off-the-grid log cabin never seemed more appealing.

Google AI Leaks Dev's Content Plans Hidden Inside Google Docs

The indie dev behind a small Steam game was baffled to learn that Google AI somehow knew private info about their game they hadn't shared online A solo indie dev behind a relatively small game has revealed that, somehow, Google's AI had gained access to and shared unrevealed information about their game. Naturally, said developer was baffled to learn this was the case, especially when the "only place where the info exists in a digital form" was within a private Google Doc. This information comes via a Reddit thread written by developer Klub Kofta Studio (thank you, GamesRadar), the solo dev behind the tower defense game Operation Octo on Steam. As Klub Kofta explained, they spotted that a user in their Discord server had been asking Google AI some questions about Operation Octo, and received some "obscure info that should not be available" in response. "Out of curiosity, I chimed in and asked the player to ask the AI about future contents in my game," continued Klub Kofta. "Somehow, the AI spitted out the exact & highly specific character name 'Vantage Tripod, which I had never mentioned to anyone. 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!" But Google has directly denied that information in your Google Drive -- or any Google service for that matter -- can be used to train its AI models. In 2023, Google's Product VP, Yulie Kwon Kim, stated that Google does "not use your Workspace data to train or improve the underlying generative AI and large language models that power Bard, Search, and other systems outside of Workspace without permission." So, if that's the case, is Klub Kofta simply mistaken? It's possible, but there's a far more likely explanation: the quote I just shared with you is woefully out of date. This isn't a case of me trying to mislead you, as Yulie Kwon Kim's statement is the most up-to-date official statement I could find published on Google's official blog. Turns out that Google has officially announced changes to its AI-training policies, though -- it's just that said information came in the form of an email. As TechCrunch covered in July, Google announced via a customer email in June that it now uses your saved history and media "to provide, develop, and improve its services (such as training generative AI models) and to protect Google, its users, and the public with the help of human reviewers." Oh, and by the way, that setting is opt-out, not opt-in, which means that you've probably been automatically signed up to share this info with Google already if you're signed up for any of its services. Here's a direct link to your Google Personalization settings if you'd like to turn that off.

Google AI leaks "scarily accurate and very specific" game detail from private documents, says indie dev: "I had never told the name to anyone"

As if the rapid proliferation of artificial intelligence wasn't already scary enough, an indie developer says it managed to leak a very specific detail from their upcoming game from private documents. Over on Reddit, solo game developer Klub Kofta Studio, whose flagship game is the aquatic tower defense title Operation Octo, shared a harrowing Discord interaction they had with a player about an upcoming update. Klub Kofta attached screenshots of the conversation in which the player asked Google's Gemini AI questions about the game and received "surprisingly obscure info that should not be available for a small indie game like mine." That in and of itself isn't shocking, but Google's AI chatbot was also talking about future content leaks gleaned from conversations on the game's Discord. This was "super impressive," said Klub Kofta, who inadvertently revealed an upcoming creature casually referred to as Tripod Fish when they drew it while playing an online party game called Gartic Phone. Discord users were able to decipher that Tripod Fish was coming to the game based on the drawing, but the exact name of the creature was never revealed publicly, according to Klub Kofta. Google Gemini was able to figure it out regardless. "Out of curiosity, I chimed in and asked the player to ask the AI about future contents in my game," explained the solo dev. "Somehow, the AI spitted out the exact & highly specific character name 'Vantage Tripod,' which I had never mentioned to anyone. 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! (The player even asked the questions in incognito)." I'm hesitant to theorize on how this all could've happened considering it's just one unverified account, but Reddit users speculated that Google automatically opts in users to allow Gemini to scan Google Drive files to train its AI. I couldn't verify that claim after some digging, but if Klub Kofta's story is true, it carries some deeply disturbing implications about online privacy in the age of AI. One Reddit user who called the story "unsettling" suggested some alternate possibilities, including scanned Discord conversations between server members, but Klub Kofta insisted the creature's name was never revealed anywhere outside of their private Google Drive. "Even the discord server members have never learned this specific name. I'm certain I'd kept it a secret, and a search history on Discord (as shown in the screenshot) indeed reveals that the keyword had never been said up until this point," said the dev. "The character name has never made it into serialized data in the game yet, so the one Google Doc where I kept future character names & descriptions was the only source of information I can think of." Forget hackers, with just a few months until GTA 6, Rockstar better be real careful about who's talking to ChatGPT.