Japanese game studio Level-5 has announced that it will rerelease corrected versions of some game trailers and artwork shown during its “Level-5 VISION 2026 II: Dreams” presentation, while also explaining how it plans to use generative AI going forward. The company said it will correct errors caused both by AI use and human errors that had been mistaken for signs of AI, with the revised videos set to be released during Tokyo Game Show 2026.
Specifically, Level-5 addressed concerns regarding key artwork for Yo-kai Watch 2: Haunted Domain. While some viewers attributed the unnatural look of the power lines and vending machines to AI use, the company explained that these elements were hand-drawn by its designers, and that some details had been intentionally simplified or omitted because they were background elements. However, there were also human errors involved.
Additionally, Level-5 explicitly stated that no AI changes or corrections were made to Nathan Adams’ face between the Nintendo Direct and Level-5 VISION 2026 II: Dreams teasers. Rather, since the deadline for the Nintendo Direct was shorter, the developers did not have enough time to complete Nathan’s model for it. The VISION 2026 II: Dreams video showcases the newest version of Nathan’s character model, thus the differences. The studio also stated that no AI was used in the video for Professor Layton and The New World of Steam.
Level-5 apologized for the distrust and confusion it caused, reflecting on having used AI too casually in some of its videos, resulting in even human errors being seen as AI artifacts. Moving forward, the studio plans to limit AI use to “efficiency-enhancing purposes” only, and to avoid direct use in promotional videos. While AI was used in the presentation, the studio stated that there is no direct use of AI in the game content of its upcoming titles.
