OpenAI may face copyright disputes due to the use of game content to train Sora

ChainCatcher news, according to TechCrunch, OpenAI's new video generation AI model Sora has been found to allegedly use game content from titles such as Super Mario and Call of Duty, as well as Twitch streamer likenesses for training. Legal experts warn that this could involve multiple copyright disputes.

Experts point out that game videos typically contain multiple layers of copyright, including the copyrights of game developers, unique content generated by players, and user-created maps. Unauthorized use of such content to train AI models may be considered infringement, and even if AI companies claim fair use, users who publish or use the generated copyrighted content may still face legal risks.

Additionally, the output content involving trademark or right of publicity issues may also increase legal challenges.