Talent agency WME is opting all of its clients out of having their likeness used in the latest version of Sora, OpenAI’s video generation tool, according to news reports.
OpenAI, the Sam Altman-run company behind chatbot ChatGPT, released a new version of Sora on September 30, which adds a variety of features to the AI video tool, including dialogue and sound effects.
It also comes with a feature called “cameos” that allows users to film a short video and audio clip of themselves, upload it to the app and have themselves appear in AI-generated video.
Sora’s ability to create hyper-realistic video of people, complete with their voice, has raised concerns in the entertainment industry that the tool could accelerate the creation of deepfakes of celebrities without their consent.
In an effort to prevent that, WME notified OpenAI that it’s opting out all its talent from being used in the new Sora app, as first reported by The Hollywood Reporter.
“Our position is that artists should have a choice in how they show up in the world and how their likeness is used and we have notified OpenAI that all WME clients be opted out of the latest Sora AI update,” said Chris Jacquemin, WME’s Chief Digital Officer, in a memo to agents.
WME represents superstars across various sectors, including fashion, film, sports, comedy, music, and more.
WME’s move is reminiscent of the letters that Sony Music Group and Warner Music Group sent to hundreds of AI companies and streaming services last year, notifying them that they are opting out from having their content used to train AI.
However, while the music companies’ letters were spurred by the European Union’s AI Act, which includes an AI opt-out clause for rightsholders, WME’s move appears to have been triggered by OpenAI’s own policies.
According to a September 29 report in the Wall Street Journal, ahead of the launch of the new Sora, OpenAI notified talent agencies and studios that their copyrighted characters will be used by the app unless the rights holders explicitly opt out.
“Our position is that artists should have a choice in how they show up in the world and how their likeness is used…”
Chris Jacquemin, WME
However, OpenAI said that Sora wouldn’t generate images of recognizable public figures unless those figures opted in.
“Our general approach has been to treat likeness and copyright distinctly,” Open AI Chief Strategy Officer Jason Kwon said, as quoted by the WSJ.
(An “opt in” option, which requires rights holders and celebrities to give explicit consent for their images or content to be used in AI, is the widely preferred approach within the entertainment industry.)
At least for now, the new Sora is generating iterations of copyrighted content. According to a separate report at The Hollywood Reporter, text prompts fed into the new Sora have returned images and characters from TV shows like South Park and Rick and Morty, movies like Dune and video games including Red Dead Redemption.
That got THR speculating on whether this will result in lawsuits from rightsholders. OpenAI is one of a number of AI developers who have been sued for copyright infringement, among them Anthropic, which is facing a lawsuit by music publishers including Universal and Concord over unauthorized use of lyrics, and Suno and Udio, two AI music generators who stand accused of mass infringement of music copyrights in training their models.
OpenAI is itself facing multiple copyright lawsuits, including a consolidated case brought by the New York Times and other news outlets, arguing that OpenAI’s ChatGPT was trained on millions of copyrighted news articles.
In an apparent response to negative feedback from rightsholders about Sora’s opt-out policy, OpenAI CEO Sam Altman penned a blog post on Saturday (October 4) that announced Sora would give rightsholders “more granular control over generation of characters, similar to the opt-in model for likeness but with additional controls.”
Altman didn’t specify what those “granular” controls will be.
“We are hearing from a lot of rightsholders who are very excited for this new kind of ‘interactive fan fiction’ and think this new kind of engagement will accrue a lot of value to them…”
Sam Altman, OpenAI
“We are hearing from a lot of rightsholders who are very excited for this new kind of ‘interactive fan fiction’ and think this new kind of engagement will accrue a lot of value to them, but want the ability to specify how their characters can be used (including not at all),” Altman wrote.
He warned that “there may be some edge cases of generations that get through that shouldn’t, and getting our stack to work well will take some iteration.”
Altman also said the company is planning to try out a revenue-sharing model on videos generated by Sora, which would see “rightsholders who want their characters generated by users” earn some of the money brought in by Sora users.
“The exact model will take some trial and error to figure out, but we plan to start very soon,” Altman wrote.Music Business Worldwide
