"Generative AI continues to plague every creative industry you care about, and despite the obvious copyright infringement and legal cases that surround it, companies like OpenAI keep training their models on licensed art. Sora, OpenAI's video generation model, launched its updated "Sora 2" model on September 30. The app trains itself on copyrighted material by default, with the burden on copyright holders to actively opt out of it ."
"By default, Sora 2 can generate animated videos of almost any copyrighted material you prompt it with because genAI companies seem to think the rules governing how copyright IP is typically handled don't apply to them. So now, companies like Nickelodeon and Nintendo will have to hit OpenAI up to tell the company to stop using their characters and iconography, rather than the other way around."
Generative AI models are being trained on licensed and copyrighted art by default, placing the burden on rights holders to opt out. OpenAI launched Sora 2 on September 30 as an updated video-generation model capable of producing animated videos of copyrighted characters when prompted. Users and media have shared examples including anime-style outputs, Pikachu, a Nazi version of SpongeBob, and Zagreus on a scooter. Companies such as Nickelodeon, Nintendo, Disney, NBC, and Warner Bros. face the new reality of requesting removal while legal actions continue against other AI firms like Midjourney. Artists have sued for scraped artwork used in training.
Read at Kotaku
Unable to calculate read time
Collection
[
|
...
]