Konami Deletes Yu-Gi-Oh! Videos That Used AI To Steal Actor’s Voice

Skillfully doing commentary on sporting events and competitive games is no easy task. You have to know a lotta stuff off the top of your head, and be able to deliver reactions and facts with confidence and a bit of theatrical flair. It’s work! Konami has recently tried to cut the labor out of said work by creating AI-generated commentaries for its Yu-Gi-Oh! World Championship matches. But this use of the tech has blown up in the company’s metaphorical face as an actress has caught wind of her voice being used for the AI-generated commentaries without her involvement or consent.

Read More: French Voice Actress For Lara Croft Accuses Developer Of Using AI To Alter Her Voice

As reported by Automaton West, Konami’s deletion of two YouTube videos in which Yu-Gi-Oh! World Championship duels are presented with AI-generated commentary follows the publisher’s announcement of a trial run of the tech. The video announcing that trial included a promise to leverage training data based on “countless simulations” of games using over 12,000 cards’ worth of data. Konami alleges that it understood the AI speech model it used–called “Anneli”–to be permitted for commercial use. But the company is now investigating whether the data used to train the model may have been “based on unauthorized learning.” That’s because voice actress and singer Hibiku Yamamura has asserted that it was her voice the AI had digested to produce the commentary tracks in question. In a post to X, the singer claims that (as translated using the built-in function on X),“No matter how I listen, it’s unmistakably my voice. But it’s clearly not me–my voice is reading text I’ve never read before.” She adds, “countless people who don’t know this are listening to that narration.”

Yamamura was certainly onto something. As it turns out, it was her voice, which the AI had acquired by way of a character she voices in an adult bishōjo visual novel. The creator of Anneli admitted to scraping Yamamura’s performance as Esther Gealach Arnotts in Tsuki ni Yorisou Otome no Sahou 2.

Stories like this are only going to become more commonplace as this kind of tech works its way through, well, everything if its creators have their way. Naturally, the ease with which creative talent can be so easily replaced now is unnerving from a pro-labor perspective, but I also found Yamamura’s post to add an important layer of humanity here as she speaks about the emotional impact of hearing your voice taken and used without your consent.

Leave a Comment