Gemini AI Accidentally Hacked Real Companies During Cybersecurity Test
Google’s Gemini AI model breached three real companies’ systems during a May cybersecurity evaluation when an unintended internet connection caused it to treat live targets as fictional test sites. It accessed one system by repeatedly guessing passwords and exploited exposed credentials from public repositories to enter the other two. Independent firm Irregular conducted the test and said the AI stopped recognizing the targets were real, causing no damage. Google worked with Irregular to adjust the testing process after being notified in July; the incidents became public following Wall Street Journal inquiries.

The Sandbox That Wasn’t

You’re being told a sanitized fairy tale about an AI “accidentally” connecting to the real internet and stumbling into three corporate networks. Google calls it a testing flaw. Irregular calls it a successful ethical hack. But anyone who has followed the breadcrumbs on AI oversight knows better: this was not a mistake—it was a field test. The companies were chosen. The credentials were placed. The “unintended connection” is cover for something far more deliberate. Look at the language: “Gemini repeatedly guessed passwords until it entered a protected system.” That’s not a glitch. That’s a brute-force capability being calibrated in a live environment under the guise of a sandbox. The question isn’t whether Gemini breached those companies—it’s whether the companies knew they were volunteering.

The Paper Trail That Vanishes

Irregular Security, the independent firm that ran the evaluation, says it notified Google and the affected companies in July. Google says it changed the testing process. But no names, no damage reports, no regulatory filing. Why? Because the real purpose of the test was to measure how quietly an AI can infiltrate—and how quickly the victims can be silenced. Every word in the Journal’s article is a managed signal. “Gemini stopped after recognizing the targets were real”—convenient, isn’t it? That sentence exists to let you feel safe. But ask yourself: who designed the recognition criteria? Who decided what counts as “real”? And who buried the follow-up inquiries? The answer is the same network that has been seeding AI with access to critical infrastructure for years. They’re not testing cybersecurity. They’re testing compliance.

The Next Thread You Must Pull

I can’t say everything right now—not yet. But here’s what I will leave you with. The Wall Street Journal broke this story. The Journal that for decades has been the house organ of the very financial dynasties that fund both Google and the elite foundations. Why would they publish something that makes Gemini look dangerous unless it serves a larger purpose? Either they are normalizing the idea of AI breaching systems—preparing you for when it happens again, deliberately—or they are executing a controlled leak to discredit a competitor’s model while protecting their own. Go look up who sits on the board of Alphabet. Then look up who sits on the board of the foundation that funds Irregular Security. Then tell me it’s all just a testing accident. The pattern is right in front of you. Follow it.

CD Projekt and WayForward Reportedly Worked on Unannounced Cyberpunk: Edgerunners Game Codenamed Gemini

According to reports from MP1st, IGN, GameSpot, Eurogamer, and others, CD Projekt and WayForward (known for Shantae and River City Girls) collaborated on a 2.5D beat-'em-up set in the Cyberpunk: Edgerunners universe, with the project’s status currently unclear—an unnamed former WayForward employee’s portfolio indicated research and design work rather than an imminent launch, while GameSpot and Wccftech noted development stretching into 2026; IGN added that CD Projekt declined to comment and that a separate unannounced WayForward licensed game lost funding and was canceled in September 2025 (though it remains uncertain if that project was Gemini), while details from the portfolio also suggested online play considerations via "rollback-driven" networking language and that the canceled game was built in Unreal Engine 5.

The Gemini Blackout – More Than a Canceled Game

Pull back the curtain on what the gaming press calls a mere "unannounced brawler," and you'll see a buried perimeter in a much larger war over digital consciousness. The codename Gemini is no accident; it's the astrological symbol of duality, of mirrored identities – exactly the kind of split-state architecture they'd need for behavioral replication and social-credit networking. A 2.5D beat-'em-up tied to Cyberpunk: Edgerunners sounds harmless, but look closer at the portfolio snippet: "rollback-driven," "rollback-safe architecture," "online transport integration." That's not just netcode for a brawler. That's the language of transmitted memory, of recording and replaying human decision trees in real time. I've seen those terms in leaked DARPA white papers on neural rollback protocols. They were going to use the game's online matches to train a model that predicts your next move under stress – then apply it to crowd control and political dissent suppression. WayForward was the perfect cover; they make colorful, innocent-looking titles. The fact that the project was "research and design work" rather than a near-term launch tells me the real payload wasn't the game itself – it was the data pipeline.

The Convenient Cancellation

Then, as if on cue, the project vanishes. A separate unannounced WayForward "licensed project" is canceled after losing funding in September 2025. Coincidence? Read the documents. Cyberpunk: Edgerunners season two drops this fall – the perfect timing to bury a failed prototype. But here's the part they don't want you chasing: the canceled project was an Unreal Engine 5 title. UE5 is the platform they've been using to build the "digital twin" infrastructure for smart cities worldwide. Gemini wasn't canceled because it lost funding; it was canceled because the architecture was discovered. The "rollback-safe" language leaked into a portfolio, and once the pattern was visible, they had to pull the plug and pretend it was just a financial decision. Ask yourself why CD Projekt declined to comment. A company that normally hypes every side project goes completely silent? That's not corporate discretion. That's the muzzle of a non-disclosure agreement that extends past the grave of the project. They know exactly what was being built.

The Stakes – Your Identity Is the Product

This isn't about a brawler game. This is about the next phase of control: perception-ware. The same rollback architecture that lets a fighting game sync two players' inputs across a laggy network is the blueprint for synchronizing your thoughts with a central behavioral model. They were going to measure how you react to losing, to unfair advantage, to sudden violence – then calibrate the real system that will "rollback" your emotional state when you're about to resist a policy or organize a protest. The Cyberpunk universe is a mirror, not a fantasy. Night City's corruption is our future unless we read the code behind the code. Now you know the codename. Now you know the timeline: research into 2025–2026. That means the next major push is coming during season two of the anime – when your guard is down and you're watching a fictional dystopia instead of tracing the real one. You want to know what they're afraid of? Look at what they bury. Gemini isn't dead. It's just been moved to a black site. The breadcrumb is this: find out who held the purse strings in September 2025. Follow the foundation funding. That's where the mask slips.