Mages, a studio widely associated with Japanese visual novels including Steins;Gate, has revealed a strikingly different project: Juggernaut, a retro-styled first-person shooter with zombies, an enormous enemy-kill objective, and an unusually detailed disclosure of generative AI involvement.

The premise appears intentionally direct. Juggernaut invokes the texture, pace, and chunky presentation of 1990s shooters, calling to mind games such as Quake and Doom. Its stated goal is simply to eliminate 5,000 enemies. That straightforward setup is paired with visuals that look to the polygonal, grimy language of early 3D action games, though the material shown has also prompted discussion over its distinctly machine-generated appearance.

What makes the announcement stand out is not merely that a known visual-novel maker is publishing a shooter. Mages’ disclosure on Juggernaut’s Steam listing says generative AI was used for 100% of the game’s source code and effects, along with 60% of its graphics and 50% of its sound. Those are exceptionally broad percentages for a commercial game disclosure, particularly for a project being presented as a completed, wishlistable release rather than a tech demonstration.

Juggernaut does not currently have a confirmed release date. It can be added to Steam wishlists, but the available information leaves major practical questions unanswered, including its final feature set, performance, pricing, and what the remaining non-generative portion of the game encompasses.

A sharp left turn for a familiar studio

Mages has long been most visible through narrative games, especially visual novels that place writing, branching presentation, character art, music, and careful pacing at the center of the experience. Steins;Gate became an international success both as a game and through adaptations in other media, helping establish the studio’s reputation well beyond its original audience.

That background does not mean Mages has never tried another format. The company released a rhythm game last year and has worked on smaller projects connected to anime properties. Even so, Juggernaut is a dramatic genre shift. A zombie FPS built around mowing down thousands of enemies has little obvious overlap with the studio’s best-known work, aside from the broad fact that both are video games and both presumably require somebody to make a menu behave itself.

The generative-AI disclosure offers a clear explanation for the speed and scale of that pivot, at least in production terms. Instead of merely employing AI for a narrowly defined task, Juggernaut’s listed use reaches into engineering, audiovisual effects, a majority of graphical material, and half its audio. The result is a title that is as much a talking point about development methods as it is about its retro shooter concept.

There is an irony to its pitch. Juggernaut is marketed around the notion of returning to the 1990s: an era remembered for id Software’s landmark shooters, aggressively experimental PC games, and technical breakthroughs born from teams pushing particular tools and design ideas. Juggernaut’s chosen route back to that aesthetic is a modern generative pipeline. It is retro by presentation, but conspicuously contemporary in how much of it was produced.

Related coverage includes Mages Reveals Juggernaut, a Retro FPS Built Extensively With Generative AI.

What the Steam disclosure says

AI disclosures are increasingly common on Steam, but their meaning can vary greatly. A game may report limited use of generated text, concept material, texture components, voice processing, or code assistance. Juggernaut’s listing is notable because it assigns percentages to several key categories rather than describing a small supporting role.

  • Source code: 100% created using generative AI models.
  • Effects: 100% created using generative AI models.
  • Graphics: 60% created using generative AI models.
  • Sound: 50% created using generative AI models.

Those figures do not, on their own, explain every stage of development. A disclosure cannot tell prospective players how much human review, revision, direction, integration, quality assurance, game design, or legal clearance occurred around the generated output. Nor does it establish whether the game will feel responsive, inventive, balanced, or technically stable. It does establish that AI use is central to the project rather than incidental.

For players, the distinction matters. A retro shooter may look simple compared with contemporary blockbuster productions, but its strengths are generally found in details that are not captured by an asset-count percentage: weapon feedback, enemy behavior, level flow, readable spaces, audio clarity, movement speed, secrets, difficulty tuning, and the rhythm of repeated encounters. The assignment to kill 5,000 enemies makes that last point especially important. Repetition can become arcade-like fun, or it can turn into a very long meeting with the same hallway.

That question of feel is also why Juggernaut will likely be compared with the classics it evokes, rather than judged only on whether it fulfills its AI disclosure. Plenty of modern independent shooters borrow the look of 1990s PC games, but their appeal usually depends on adding a distinct mechanical hook, memorable level design, or a voice of their own. The early material positions Juggernaut close to familiar landmarks, which raises the challenge of explaining what it offers beyond recognizability.

Reaction centers on both style and scale

Early responses from English-speaking fans have been strongly negative. Some reactions focus on the visual presentation, while others are aimed at the scope of the AI usage itself. For an audience that connects Mages with authored, character-focused storytelling, a project whose disclosure lists generated code for the entire game is bound to read as a fundamental change in priorities.

The response also reflects a broader concern that “retro” can become an excuse for work that resembles old games without demonstrating the intentional craft that made those games endure. Low-poly models, dark stone corridors, hostile monsters, and noisy action are easy signifiers. Creating the tension, responsiveness, identity, and replay value associated with the genre is a much less automatic job.

It is worth separating what is known from what cannot yet be concluded. Juggernaut’s disclosed percentages are specific, and its wishlist status is real. A release date has not been announced. There is not enough available information to determine the game’s final quality, how its AI-assisted materials will function in motion, or whether it will develop an identity more distinctive than its first impression. Treating the disclosure as a guarantee of either disaster or success would go beyond the available facts.

Japan’s games business is actively exploring generative AI

Mages is not operating in a vacuum. A report from last year found that 51% of Japanese game studios were experimenting with generative AI. Other companies, including Level-5, have publicly discussed or shown ways generative assets can be used in development. That broader interest helps explain why AI has become an industry subject rather than a fringe curiosity.

Experimentation, however, covers a vast range of practices. Using a tool during ideation is different from using it for a single type of asset; both differ from placing it at the center of a project’s code, visual effects, art, and audio pipeline. Juggernaut is unusual because the stated percentages make the scale visible. Even in a business increasingly willing to investigate the technology, it is rare to see a game identify generative models as the origin of all its source code and effects while also accounting for large portions of art and sound.

That transparency may become the most consequential part of the reveal. Clear disclosures allow players to make purchasing decisions based on their own standards and give developers, artists, and publishers a concrete case to debate. They also avoid the uncertainty that arises when generative tools are suspected but never acknowledged. In Juggernaut’s case, there is no need to guess whether AI participated in production; the listed numbers make its importance unmistakable.

The larger debate will not be settled by one zombie shooter. Questions about authorship, labor, reliability, originality, training data, rights management, and the value of human-made creative work will persist well beyond Juggernaut. But the game has put many of those questions into a compact, unusually legible package: a studio famed for visual novels, a throwback FPS, 5,000 enemies, and a development disclosure broad enough to become the story itself.

For now, Juggernaut remains an announced PC game without a release date. Its Steam page gives potential players the option to wishlist it, while its generative-AI breakdown gives them an unusually direct basis for deciding whether they want to follow it at all. The industry’s wider interest in unconventional production tools continues, as does interest in PC projects that preserve or reinterpret older traditions, as seen in ongoing efforts to improve the future of classic PC games. Juggernaut approaches that past from a very different direction—and players will ultimately decide whether that approach has any staying power.