ByteDance Japan announced on February 26 that it has made service adjustments to its latest released video generation AI model, Seedance 2.0, aiming to strictly control potential rights infringement risks in generated content.
This adjustment comes in response to reports that the model, after its launch, was capable of generating unauthorized videos featuring well-known anime IP characters such as Disney and "Ultraman." On the 24th, the Japanese Minister for AI Strategy, Kiyomi Onoda, clearly stated at a press conference that the government has requested relevant operators to take corrective actions and reminded AI users that they may also face compensation or legal liability.
As a major product in ByteDance's multimodal field, Seedance 2.0 adopts a unified audio-visual joint generation architecture, supporting the deep integration input of four modalities: text, images, audio, and video, and possessing strong multimodal content reference and editing capabilities. However, technological breakthroughs come with copyright challenges. Walt Disney Company sent a lawyer's letter to ByteDance earlier this month, accusing the model of using its copyrighted works without permission during training and development, and demanding an immediate halt to the infringement.
This incident highlights the intense conflict between the pursuit of multimodal realism in generative AI and the boundaries of copyright protection. As high-capability models like Seedance 2.0 become more widespread, investments by large model developers in data compliance and content filtering mechanisms are gradually shifting from technical add-ons to core criteria determining product viability.
