The rise of generative AI has made the already grim reality of the modern games industry even more complicated, threatening to accelerate layoffs, climate change, and theft from artists all in one fell swoop. That’s led to rampant accusations (only some of which are accurate) of developers using genAI to make their games. Nintendo is the latest company caught in the crosshairs, thanks to a major part of its Switch 2 reveal.
When Nintendo showed off the Switch 2 in April, its presentation put the spotlight on Mario Kart World, a semi-open world version of the company’s popular racing game. But some players say images shown in a Nintendo Treehouse stream of the game show telltale signs of being made with generative AI, a claim that Nintendo has denied.
“AI-generated images were not used in the development of Mario Kart World,” Nintendo told Eurogamer after rumors began swirling.
Nintendo did not immediately respond to Inverse’s request for comment.
The controversy began after some viewers of a Mario Kart World stream from last month began to voice suspicion over some of the art on display. Some areas of the game are decorated with billboards showing images that viewers say have a striking resemblance to the output of AI image generators.
Looking at the widely shared images up-close, there’s no denying that something looks a bit off about them. In one billboard, the supports of a bridge are made of wobbly lines, with inexplicable tangents and background objects cluttering up the frame. In another, a construction site takes on an almost M.C. Escher look with beams that don’t seem to connect in any reasonable way. Another suspicious image shows a car with an oddly tall cabin, with similarly bizarre lines that don’t seem to serve any purpose in the image. These elements tend to be hallmarks of generative AI.
Still, a Nintendo spokesperson says that AI wasn’t used in their creation, and there’s no way to objectively prove whether that’s true or not. Some developers and players find it more excusable to use generative AI to create temporary placeholder images and replace them with actual art for the final game, but Nintendo’s assertion suggests that the technology wasn’t used in the development process at all.
That lines up with previous comments from Nintendo’s Shigeru Miyamoto, who has disavowed the use of generative AI entirely. In a 2024 interview with The New York Times, Miyamoto spoke about Nintendo’s refusal to follow the trend toward more photorealistic graphics that dominates much of the games industry.
“It might seem like we are just going the opposite direction for the sake of going in the opposite direction, but it really is trying to find what makes Nintendo special,” he said. “There is a lot of talk about AI, for example. When that happens, everyone starts to go in the same direction, but that is where Nintendo would rather go in a different direction.”
Nintendo president Shuntaro Furukawa acknowledged that generative AI can conflict with intellectual property rights during a shareholder call, though he also said that it “can be used in creative ways.” They’re far from a promise that Nintendo will never use generative AI, but Miyamoto’s unprompted comment and Furukawa’s reticence do suggest at the very least a desire to avoid it.
The questionable Mario Kart World art could just be slightly wonky human-made placeholders intended to be cleaned up before release. Given the myriad ethical issues surrounding genAI, it’s disappointing that so few developers have been willing to step up and affirmatively say that they reject its use on those grounds. But even if they’re not willing to do that, the outcry over the mere potential that Nintendo is using the technology should be a wakeup call. The specter of generative AI has made players paranoid over any hint that it’s wormed its way into another game, and developers would be wise to make their rejection of it loud and clear.
