// Workers AI · dad joke modeWhat did the seedance goer say? You have to weed out the bad dancers.

Seedance
Developer	ByteDance
Release	June 2025 (2025-06)
Stable release	Seedance 2.5 July 31, 2026 (2026-07-31)
Type	Text-to-video model
Website	seed.bytedance.com/en/seedance2_5

Seedance is an AI text-to-video model created by ByteDance and first released in 2025. Seedance 2.0 quickly went viral for its ability to create short video clips featuring photorealistic scenes of famous actors and characters. Its level of realism caused fascination, particularly in China, but also led to concerns about widespread copyright infringement, potential misuse in spreading disinformation and career impact on human performers due to its ability to replicate Hollywood-style film production. The latest version, Seedance 2.5, was released in July 2026.

History

Seedance 1.0 and 1.5 Pro

Seedance was first released in June 2025. ByteDance's Seed research team published a technical report on Seedance 1.0 on June 11, 2025, describing a model that generates 1080p video from text and image prompts, including clips with multiple shots.

In December 2025, ByteDance released Seedance 1.5 Pro, which generates audio and video together in a single model. It added lip-sync in multiple languages and dialects, sound effects and background music, along with more complex camera movements such as the dolly zoom.

Seedance 2.0

Seedance 2.0 was released on February 12, 2026, with the ability to generate clips of up to 15 seconds. Shortly after it was released, realistic clips based on real actors, TV shows and films went viral across the Internet. Clips included Friends characters reimagined as otters, a fight between Brad Pitt and Tom Cruise, and Will Smith engaged in battle with a red-eyed spaghetti monster, creating a nemesis out of his history of eating spaghetti.

Copyright disputes

Seedance was denounced by the Motion Picture Association for copyright infringement quickly after release. On February 13, 2026, The Walt Disney Company sent ByteDance a cease and desist letter alleging that the model had been trained with Disney works without any compensation. Paramount Skydance accused the company of engaging in "blatant infringement" of its intellectual property including Star Trek, South Park, and Dora the Explorer. SAG-AFTRA also condemned the tool, citing "the unauthorized use of our members' voices and likenesses". On February 16, 2026, ByteDance said that it "respects intellectual property rights" and had "heard the concerns regarding Seedance 2.0", and that it was taking steps to strengthen its safeguards against the unauthorized use of intellectual property and likeness.

On February 17, 2026, Netflix sent ByteDance a cease-and-desist letter threatening "immediate litigation" and describing Seedance as "a high-speed piracy engine". The letter cited Stranger Things, Bridgerton, Squid Game and KPop Demon Hunters. On March 16, 2026, United States Senators Marsha Blackburn and Peter Welch wrote a letter to ByteDance CEO Liang Rubo asking him to shut down Seedance and implement safeguards, stating that it is "the most glaring example of copyright infringement from a ByteDance product to date, and you must immediately shut down Seedance and implement meaningful safeguards to prevent further infringing outputs."

On March 26, 2026, ByteDance began rolling out Seedance 2.0 outside of China through its CapCut video editor under the name Dreamina Seedance 2.0. The initial markets were Brazil, Indonesia, Malaysia, Mexico, the Philippines, Thailand and Vietnam. The United States was not included. The CapCut version does not generate videos from images or videos containing real faces, blocks unauthorized use of intellectual property, and adds an invisible watermark to its output.

Seedance 2.5

Seedance 2.5 was released on July 31, 2026 through ByteDance's Jimeng AI and Doubao apps. According to ByteDance, the model generates audio and video together in clips of up to 30 seconds in a single pass, which can be extended over multiple rounds into videos lasting several minutes long. A single request can include up to 30 images, 10 video clips and 10 audio clips as references. The model also supports editing at specific timestamps, so users can change part of a clip, such as a character's appearance, without regenerating the whole video.

Reception

Seedance 2.0 caused alarm in Hollywood. After the clip of Pitt and Cruise fighting went viral, screenwriter Rhett Reese, co-writer of Deadpool & Wolverine and Zombieland, wrote on social media: "I hate to say it. It's likely over for us", further stating that "In next to no time, one person is going to be able to sit at a computer and create a movie indistinguishable from what Hollywood now releases." Some doubted how the clip had been made, software developer Aron Peterson questioned if the scene was purely AI generated or if ByteDance was using a video-to-video workflow behind the scenes; Seedance's website provided several examples of reference videos, including a fight that was shot with stuntmen against a green screen backdrop.

Reaction in China was more positive. The general public was largely enthusiastic about Seedance, and the Chinese entertainment industry has generally been more open to AI than Hollywood. Jia Zhangke, a Chinese film director, posted a video which remade classic scenes from his films that he said was created with ByteDance's Doubao chatbot. Writing on Weibo, he stated that "I don't worry about whether technology will replace movies," and that "What actually matters is how the technology is used by people."

See also

• Sora (text-to-video model)
• Google Veo
• LTX (text-to-video model)
• Runway Gen