When a game refuses to revive a voice, AI stops being just a tool
A torchlit Darkest Dungeon-style narrator booth left empty, with a vintage studio microphone and manuscript pages as a symbol of absence over imitation.📷 AI-generated image / TECH&SPACE
- ★AI voice rejected
- ★Performance protected
- ★Consent is not enough
When a studio turns down a tool it could have used, that is usually more revealing than the tool itself. According to PC Gamer's report, Red Hook co-founder Chris Bourassa refused to generate the voice of late narrator Wayne June even though June had granted permission in one of his final emails to train an AI system on his voice.
The studio also donated to June's family, which separates the decision from any narrow cost-benefit logic.
That distinction matters because this is not the standard "rights versus innovation" argument. The right, at least as described in the source, was there. What Red Hook rejected was not legal access but the idea that a performance central to Darkest Dungeon should be extended through a synthetic continuation as if it were just another technical asset to maintain after the actor's death. When Bourassa says he would not want to erode a timeless performance, the point is not sentimental branding.
The point is that the voice in this game is not a neutral function. It is part of the work's identity.
Red Hook Studios is taking a stance that may sound old-fashioned to some, but it is actually editorially exact. If the original performance matters enough that players recognize it immediately, then endlessly reconstructing it can produce the opposite of preservation. Instead of honoring a legacy, it turns that legacy into a service layer. That is especially sensitive in games whose mood depends on a singular narration style, cadence, and vocal texture rather than just the text being read aloud.
The Darkest Dungeon studio decided its late narrator's signature performance should not be turned into a synthetic substitute, even with his consent.
A close, moody studio scene showing waveform screens powered down beside annotated narration pages, emphasizing the decision to stop rather than synthesize.📷 AI-generated image / TECH&SPACE
That is why this story travels beyond a single franchise. AI voice cloning is often marketed as a compromise solution: you have consent, you have training data, you have a model, so why not continue. But creative industries do not fracture only along the line of permission. They also fracture along the line of substitution. The fact that something can be replicated does not answer whether it should be, especially when the performer is gone and the work that made them memorable already exists in a complete, public form.
This is where Red Hook's choice carries weight. It is not an anti-AI manifesto, and it is not technophobic theater. It is a specific judgment that an imitation of Wayne June could weaken what made the original work matter. For the studio, it is apparently better to live with absence than to manufacture a presence that is similar enough to satisfy a production pipeline but not real enough to carry the same cultural force.
For the games industry, that is an uncomfortable but useful lesson. AI tools will increasingly offer continuation exactly where creators may need to acknowledge a break instead. And audiences will need to separate two ideas that are often blurred on purpose: respect for someone's work and extraction of that work's recognizability. In Wayne June's case, Red Hook chose the harder path, but also the cleaner editorial answer.

