影视飓风视频被字节训练？Tim直言Seedance2.0模型恐怖

Core Insights - The article discusses a shocking discovery made by a user, Tim, while using the Seedance 2.0 platform, which can generate a highly accurate voice that resembles the user based solely on a facial photo upload [5][8] - The technology demonstrated by Seedance 2.0 raises concerns about data privacy and the implications of AI-generated content, as it can recreate not only voices but also visual elements without explicit user consent [5][8] Group 1 - Tim discovered that Seedance 2.0 could generate a voice that closely matches his own using only a facial photo, without any prior audio or information uploaded [5] - The system was able to create a video that included perspectives not visible in the uploaded photo, indicating advanced capabilities in understanding and recreating environments [5] - This revelation suggests that the content from the film team "影视飓风" has been utilized by ByteDance for training the Seedance 2.0 model, raising ethical questions about data usage [5] Group 2 - Tim expressed uncertainty about whether the platform's user agreement contains hidden data authorization clauses, noting that he has never been compensated or contacted for permission [8] - Testing with other creators' materials showed that Seedance 2.0 has a high fidelity in replicating individual likenesses, further emphasizing the technology's potential impact [8] - The article warns that if an individual's data is fully integrated into an AI dataset, it could lead to indistinguishable simulations of their appearance and voice, posing significant challenges for authenticity recognition by family and friends [8]