AI video games have truly entered god mode. ByteDance, the company that brought you TikTok and CapCut, released Seedance 2.0 last week, and it’s a real mess, in a good way. This amazing tool lets you mash text, photos, clips, and audio to create ultra-realistic videos that look like Hollywood blockbusters. No more spooky uncanny valley vibes with natural movements, slapping lip sync, and on-point lighting.
In just a few days, creators flooded their feeds with deepfake-style clips featuring various celebrities. Imagine Kanye West singing classical Chinese songs in an ancient palace (after barging into the room with his ex-wife Kim Kardashian, surrounded by Travis Scott and iShowSpeed clones). Or Justin Bieber in front of the Hong Kong skyline and Jet Li vs. Jackie Chan in an old movie setting. Or Stephen Chow fooling around with NBA star Kawhi Leonard. Fight Back to School 2. Other clips generated included Disney’s Spider-Man, Iron Man, and Grogu, aka “Baby Yoda.”
Directed by Irish film director Ruairi Robinson. last days on marsposted a 15-second clip of Brad Pitt and Tom Cruise throwing punches at each other in an intense scene on a rooftop. The fabric will move to the right. Their hair physics tracks naturally. Their facial expressions are in perfect sync with their dialogue. The lighting also matches the scene. This is no longer deepfake junk, but Hollywood-level output. All of this can be done with just a “two-line prompt” and render time is probably less than a minute. Not to mention, Seedance 2.0 is currently available to try for free, and ByteDance’s complete AI video suite can be accessed starting at $18/month. Rhett Reese, co-author deadpool & wolverineshared a clip of Robinson on X and said, “I hate to say it. It’s probably over for us.”
It will help you understand how easy Seedance 2.0 is to operate. This tool is multimodal magic. Simply feed up to 12 assets, including up to 9 images, up to 3 short videos (less than 15 seconds), and up to 3 audio files (less than 15 seconds), and the AI will add text prompts that tell you how to use those assets as hard references. Want to see a video of Drake performing in a neon dystopia? Upload a photo of him, a beat drop, and describe the vibe. BOOM: Realistic movement with physics, fabrics that flow correctly, and faces that emote perfectly. Synchronize audio and visuals. Sync lip-synced dialogue, sound effects to match your actions, and even choreography to your uploaded tracks.
We mere mortals will laugh out loud at Seedance 2.0, but the pros will eat it up given the cost savings that come with such quality, consistency, and speed. 9:16 for TikTok or 16:9 for YouTube? The end. Scene enhancements, character swaps, and frame tweaks are all done post-generation. ByteDance, through its Seed site, calls it “the most comprehensive multimodal content reference,” surpassing benchmarks such as text-to-video conversion and image-to-video conversion.
Understandably, Hollywood is very concerned about this upgraded GenAI tool. Within days of Seedance 2.0’s launch, Disney and Paramount sent ByteDance a cease-and-desist letter alleging massive copyright and publicity rights infringement. The Motion Picture Association of America (MPA) also accused ByteDance of “massive misappropriation of U.S. copyrighted material” in a statement.
What are their claims? These AI videos are training on protected content. Jet Li deepfake? Probably trained based on his actual filmography. Jackie Chan video? His movies, his portraits, his signature movements. Kanye West deepfake? His image, voice and personality are all legally protected intellectual property. You can’t run AI on that content and generate infinite variations without compensation.
But it gets even trickier. Tom Cruise and Brad Pitt’s deepfakes touched off another legal nerve, as both actors have massive franchises. A cruise literally means mission: impossiblea Paramount treasure. Pitt has extensive IP associated with his image through various studios. When you produce a convincing video where they may be doing something they’ve never done, wearing something they’ve never worn, or saying something they’ve never said, you’re undermining your right to publicity, your legal right to control how your image or likeness is used for commercial purposes.
Wild card? No one knows exactly what Seedance 2.0 was trained on. ByteDance does not fully disclose training data. However, given their ability to perfectly recreate the likeness and mannerisms of certain celebrities, the studio assumes that the models have captured a large amount of footage, photography, and audio recordings from libraries and the wider Internet.
Disney is feeling particular pressure because of Marvel, Star Wars, and decades of legacy content. If Seedance 2.0 is able to generate convincing deepfakes of characters associated with its IP, it poses a threat to their entire business model. That said, OpenAI signed a deal in December that allows it to feature Disney characters on its Sora video generator, so it’s just a matter of money.
What is undeniable is the fact that advances like Seedance 2.0 are democratizing content creation in ways that scare big companies. Anyone with a laptop or mobile phone can now produce Hollywood-quality videos. There is no crew. There is no budget. There is no gatekeeping. Get unlimited creative variations without reshoots. This is truly a world-changer for independent filmmakers, TikTok creators, small businesses, and literally anyone looking to create video content without shedding money.
But what about the other side? Consent and ethics are disappearing. These celebrities never consented to their likenesses being used in this way. Kanye didn’t allow his images to be used in deepfakes (well, maybe — knowing Kanye, he might actually be here to cause confusion, but that’s beside the point). The broader question is whether AI should be allowed to clone people’s identities without their permission.
Legally, the studio is fighting for a “no.” Creatively, the internet is fighting for “who cares, this is a fire”. And technically, ByteDance hasn’t stopped it. In fact, technology isn’t going away. Other companies are also building their own versions. The genie is out of the bottle and cannot be pushed back.
The future will probably be some kind of hybrid. That means a celebrity consent protocol that requires creators to flag when they use celebrities, licensing agreements that compensate studios and talent, and a clear legal framework on what constitutes fair use in AI generation.
For now? Enjoy the chaos. Download Jet Li deepfake. Kanye duet video. Remix Justin Bieber clips. Because this could be a very short period of time during which we can create something before the lawyers close the door for good.
Cover image from YouTube.
