Square Enix and Studio Ghibli Want OpenAI to Stop Ripping Off Their Work

Square Enix and Studio Ghibli Want OpenAI to Stop Ripping Off Their Work

OpenAI, developer of ChatGPT and Sora, has received a written request from the Japanese anti-piracy group Content Overseas Distribution Association to stop using their members’ content to train their recent video generator Sora 2 without explicit permission. CODA’s members span the entertainment industry, including game publishers like Cygames (UmaMusume: Pretty Derby) and Square Enix (Forspoken), as well as companies like Studio Ghibli (Ponyo) and Universal Music (Taylor Swift’s The Life of a Showgirl).

Launched on September 30, Sora 2 and the accompanying Sora app is OpenAI’s flagship video generating technology, trained on untold amounts of footage. Using the app is similar to TikTok, except for the fact that nothing shown is real or was created by actual human beings. Checking the front page shows you a clip of a dog in a tutu pirouetting as recorded by a security camera and a clip of someone feeding a grape a grape, all based on a prompt and drawing upon Sora’s swaths of video data.

You may recall the flood of “Ghibli” avatars and images back in the Spring off the back of OpenAI’s GPT-4 image generation. This rightfully was pushed back on by those with taste, and evidently Ghibli and its fellow CODA members took notice, issuing their letter to OpenAI on October 28. In addition to requesting that Sora not be trained on its members’ content without permission, the letter also asks the company to “respond sincerely” to claims of copyright infringement from its members.

So far, OpenAI CEO Sam Altman has responded on his personal blog, saying that the company is going to make changes in response to feedback from “rightsholders.” First they intend to give “rightsholders more granular control over generation of characters, similar to the opt-in model for likeness but with additional controls” while claiming that said rightholders are excited about the “interactive fan-fiction” OpenAI’s regurgitation allows for. Second, Altman says that OpenAI will “try” to share some of their revenue with those they have built their business on. 

CODA is primarily an anti-piracy group, similar to the ESA in the games world. And although it’s tempting to praise anybody going against the AI industrial complex, nothing’s black and white. CODA files complaints to governments regarding piracy sites making anime widely available for free, and they don’t have any room for nuance; they recently took down 16 sites in Brazil that provided Brazilian Portuguese subtitles to their viewers. CODA’s goal is to protect the IP rights of their members and ensure that consumers interested in reading, watching or playing media from the companies go through whatever means of distribution the companies approve of, even if it’s not available in their language. If the art isn’t legally available with Brazilian Portuguese subtitles, then people simply should accept it, CODA argues. So even though these companies, who retain the copyright to their artists’ work for decades, might be acting in most people’s interests when it comes to AI, it’s important to remember that they ultimately serve their own interests above all else.

 
Join the discussion...