What happened

The Motion Picture Association (MPA), which represents major U.S. film studios, publicly accused ByteDance’s Seedance 2.0 of “unauthorised use of US copyrighted works on a massive scale.”

Seedance 2.0 can generate highly realistic, cinematic video clips, and users quickly began posting short scenes that appeared to feature Hollywood stars and recognizable franchise imagery on social media.

The MPA says that in just a single day of public testing, the service produced large volumes of clips that drew on protected U.S. films and characters without permission, potentially harming rights‑holders’ ability to control and monetise their work, and potentially putting them out of business.

In one instance, Oscar-nominated Irish filmmaker Ruairí Robinson shared an ultra-realistic AI-generated video on social media featuring the likenesses of Tom Cruise and Brad Pitt having a fistfight on a Los Angeles overpass. 

Disney and Paramount also sent ByteDance cease-and-desist letters alleging Seedance 2.0 was trained on protected characters, calling the AI’s use of intellectual property “blatant infringement.” And Hollywood unions condemned the tool for unauthorised use of actors’ likenesses and voices.

While Seedance is currently in limited test release in China, the resulting AI videos circulate globally online, so the alleged harm extends beyond Chinese users and could impact U.S. and international audiences and markets.

Why it happened

Seedance 2.0 is designed to take short text prompts and turn them into polished, story‑driven video, which typically requires training on vast amounts of existing video content, including popular films and TV shows.

The MPA argues that ByteDance launched the system without “adequate protections against infringement,” meaning users can easily prompt it to mimic specific copyrighted works or stars, with no effective safeguards or licensing in place.

Cross‑border enforcement is difficult: U.S. copyright law protects the studios’ works, but Seedance is developed and tested in China, creating a gap between where the AI is run and where the rights are held.

The incident also reflects broader transparency and accountability problems in generative AI: companies rarely disclose the specific training datasets or the mechanisms they use to prevent output that closely copies protected content.