ByteDance rectifies and officially launches Seedance 2.0 on CapCut, but with restrictions for copyrighted images

ByteDance has rectified and finally launched its model Dreamina Seedance 2.0 for the generation of videos with artificial intelligence (AI) through CapCuthowever, he has done so with restrictions to protect copyrighted content and the ban on ‘deepfakes’.

On March 16, the company announced the decision to suspend the global launch of Seedance 2.0 after disputes with representatives of the film sector in Hollywood for creating realistic image videos with no limits on the use of copyrighted content.

Now, the company has resumed its initial plans and has officially launched Seedance 2.0 adding restrictions and limitations to protect intellectual property, in addition to prohibiting ‘deepfakes’ to prevent images such as the viral AI-generated video of Brad Pitt and Tom Cruise fighting in a ruined building from being repeated.

Concretamente, ByteDance has vetoed the possibility of generating videos from images that contain real faces. Likewise, it has implemented advanced technologies to block unauthorized content with rights usage, including proactive monitoring methods, in-app reporting, and direct collaboration with creators and rights holders.

The technology company has also introduced a “invisible watermark” to identify videos generated with Seedance 2.0. This flag will allow you to “take action” in cases where content violates their guidelines, even if it has been shared or modified outside of CapCut.

However, the company has clarified in a statement on its website that these measures are based on others that were already integrated into CapCut to advance the transparency of AI and the prevention of content that violates its policies. In this way, CapCut labels AI-generated content with a visible watermark y C2PA credentials to check if it has been done with AI.

For the moment, the launch of Dreamina Seedance 2.0 will take place gradual way for paying CapCut users at Indonesia, Philippines, Thailand, Vietnam, Malaysia, Brazil and Mexico. In addition, ByteDance has advanced that They will be in contact with experts and creative communities as they move forward with expansion to “continuously learn and improve on the ways this model can empower human creativity.”

The multimodal image and video generation model Seedance 2.0 was presented in China last February, and stands out for introducing quality improvements compared to its predecessor when generating professional resultscon Very realistic visual effects and stereo sound dual channel.