The excitement around Let It Die: Inferno evaporated almost instantly once fans noticed a small but explosive sentence added to its Steam page: the team admitted that some in-game voices, music, and visuals had been generated using AI before being edited by hand. Following the backlash, Supertrick has now published a detailed breakdown of what was made with AI, insisting that the tools “respect copyright laws” and are “not modeled after any human performers.”

The surprise sequel to Suda51’s 2016 roguelike, Let It Die: Inferno found itself under fire last week when a newly added AI-content disclaimer raised far more questions than it answered. After a flood of criticism, Supertrick released a lengthy explanation meant to calm the situation. Ironically, the scope of the AI usage is so limited that many now wonder why the studio bothered with it at all.

For background art, Supertrick says the team first created their own concepts and text descriptions that established the world’s setting. Only after that did they run an AI tool—one they claim follows copyright rules—to generate “rough base images,” which were then painted over, refined, and fully integrated by human artists. According to the studio, this method was applied only to a handful of posters, some InfoCast inserts, and a few images included in the game’s readable materials.

The broader AI debate has also pulled other studios into the conversation. Swery and Suda51’s Hotel Barcelona recently issued a similar apology for using AI assets, while the team behind Demonschool went as far as saying they would “rather cut off their own arms” than rely on GenAI professionally—rebuking remarks from a Nexon executive who claimed that “every studio is using AI now.”

On the subject of voices, Supertrick says only three characters were voiced with AI: Mom, Goz, and Mez. The studio explains that Mom is literally an AI machine in the story, while Goz and Mez are mysterious lifeforms whose artificial voices were chosen intentionally. Supertrick stresses that none of these voices were generated from real actors, nor were they modeled after any human speech patterns, avoiding copyright issues entirely.

Music saw similarly minor usage. Only one track—Select Iron Perch BGM—was created using an “AI-based music editor” to output stems. Even then, the devs claim most of those stems were rebuilt manually afterward, with only a few edits made to the original AI output.

In the end, the studio’s explanation confirms that the AI footprint is much smaller than players initially feared. Still, one question lingers in the air: did all of this truly save development time, or did it just spark a controversy that wasn’t worth the trouble?

Source: PC Gamer