All copyright holders will soon be able to define how, and if, their work is recreated by Sora 2. Since its release last Tuesday, OpenAI’s new video generator has been used to create videos featuring popular characters like SpongeBob SquarePants, Pikachu, and Mario.
While many of the videos have been humorous, some have been offensive, such as one featuring SpongeBob as a dictator giving a speech. Sam Altman, OpenAI’s CEO, has announced that he wants to give copyright holders more “granular control.”
In his blog post announcing Sora 2’s first round of updates, Altman says he aims to make the system for copyright holders similar to the opt-in model it employs for regenerating people’s likenesses. Sora 2 will not generate videos of recognisable public figures without their consent, and individuals can control who can view or generate AI “cameos” of themselves.
Altman claims that rightsholders may benefit from Sora 2’s ‘interactive fan fiction’
Naturally, Altman does not want all rightsholders to immediately opt out of allowing Sora 2 to reproduce their work; playing around with favourite characters has proven one of its most popular uses so far, especially with Japanese content, leading it to fly up to the top of the App Store charts.
He claims that many rightsholders 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).”
Just before the model and its accompanying Sora app were launched, reports indicated that OpenAI was contacting copyright holders to give them the option to exclude their IP from Sora 2. It also accepts takedown requests through a dedicated copyright disputes form, but will not act on blanket opt-out requests covering all of an artist’s or studio’s works.
There is evidence that these are being honoured: Disney has opted out, according to Reuters, and users have reported being unable to generate Sora 2 videos featuring Spider-Man and Darth Vader. However, many characters remain available in Sora 2, and Altman wants to ensure that all rightsholders, whether or not they had the chance to opt out before launch, can still withdraw their consent if they want to.
OpenAI does not want any more lawsuits; it already has plenty as a result of ChatGPT’s training data and has signed several licensing deals with publishers to stem the tide. The change of heart from an opt-out to an opt-in system may reflect not only a bid to secure cooperation from animation studios for Sora 2’s outputs, but also an attempt to earn forgiveness for any unauthorised training on their content that may or may not have occurred.
Rightsholders who don’t opt out should eventually be paid
Altman wants to make a “compelling” case for not opting out, and part of that is revenue sharing. He has announced that OpenAI plans to employ methods that will enable the firm to earn money from video generation, and “some” of this revenue will be given to rightsholders who permit their characters to be generated by Sora 2.
OpenAI said upon Sora 2’s release that it will eventually offer users the option to pay to generate an additional video if demand exceeds available compute resources. “Our hope is that the new kind of engagement is even more valuable than the revenue share,” he wrote. “But of course we … want both to be valuable.”
Altman also caveated that some copyrighted material may be generated by Sora 2 even if the rightsholder has opted out, and that the revenue-sharing model “will take some trial and error” to get right. The AI video generator will have a “very high rate of change,” so that these kinks can be ironed out.
OpenAI has been valued at $500 billion, making it the most valuable startup in the world. However, its potential shift to a for-profit model is still being debated in court.
The post OpenAI Is Going To Let Copyright Holders Opt Out of Sora 2 appeared first on eWEEK.
No Responses