After spooking Hollywood, ByteDance will tweak safeguards on new AI model

TikTok creator ByteDance says that it is moving to amended safeguards connected its caller AI video generator aft Disney, Paramount, and Hollywood waste and acquisition groups accused the instrumentality of violating copyright protections. Concerns were raised aft hyperrealistic videos generated by the Seedance 2.0 model went viral past week, pinch the likeness of actors, specified arsenic Tom Cruise and Brad Pitt, and characters from Dragon Ball Z, Family Guy, and Pokémon.

“ByteDance respects intelligence spot authorities and we person heard the concerns regarding Seedance 2.0,” a ByteDance spokesperson said successful a connection shared by CNBC. “We are taking steps to fortify existent safeguards arsenic we activity to forestall the unauthorized usage of intelligence spot and likeness by users.”

ByteDance’s consequence comes aft receiving complaints from awesome studios. In a cease and desist missive sent connected Friday, Disney accused ByteDance of “hijacking” its protected characters by “reproducing, distributing, and creating derivative works” that characteristic them. Deadline reports that Paramount Skydance followed pinch its ain cease and desist letter, demanding that ByteDance region each infringing instances of Paramount contented and forestall it from being generated successful the future.

Hollywood waste and acquisition groups person besides weighed in, including the Motion Picture Association (MPA) and SAG-AFTRA talent union. MPA CEO Charles Rivkin issued a statement accusing ByteDance of engaging successful copyright infringement astatine a “massive scale,” and disregarding well-established laws that underpin “millions of American jobs.” SAG-AFTRA said that its members’ voices and likenesses person been utilized by Seedance 2.0 without authorization.

“This is unacceptable and undercuts the expertise of quality talent to gain a livelihood,” SAG-AFTRA said successful a connection to AP News. “Seedance 2.0 disregards law, ethics, manufacture standards and basal principles of consent. Responsible AI improvement demands responsibility, and that is nonexistent here.”