Seedance 2.0：当AI接管好莱坞，狂欢背后的版权风暴

Speaking of the most jaw-dropping developments in the tech and film industries recently, one absolutely cannot avoid Seedance 2.0, just released by ByteDance.

The advent of this AI video generation tool is like gamers turning on "God Mode," directly razing the technical threshold of video production to the ground.

It allows users to mix text, images, video clips, and audio together to generate hyper-realistic videos with the quality of Hollywood blockbusters.

The "Uncanny Valley" effect—those stiff movements and eerie lighting we often saw in AI videos before—has almost completely disappeared in Seedance 2.0.

Replacing them are fluid and natural body movements, precise lip-syncing, and impeccable scene lighting.

Within just these few days, social media recommendation feeds have been flooded with all kinds of AI videos that are hard to distinguish from reality.

For example, you might have already scrolled past the absurd scene of Kanye West passionately singing old Chinese songs in an antique-style Oriental palace.

Or seen Justin Bieber standing in front of Hong Kong's Victoria Harbour, or even Jet Li and Jackie Chan in a peak showdown within a retro movie setting.

Even more interestingly, someone generated footage of Stephen Chow and NBA star Kawhi Leonard interacting in scenes from "Fight Back to School II."

These cross-time, cross-dimensional collaborations that originally existed only in fantasy can now become reality with just a few prompts.

A 15-second video released by Irish director Ruairi Robinson has further ignited public opinion.

In the video, Hollywood superstars Brad Pitt and Tom Cruise engage in a fierce fight on a rooftop.

Whether it's the physical effects of clothes swaying in the wind, the movement of hair, or even the synchronization rate of facial expressions and dialogue, everything has reached an astonishing professional level.

And all of this was rendered in less than a minute based on merely two simple lines of prompts.

After watching this video, Rhett Reese, co-writer of "Deadpool & Wolverine," even lamented pessimistically on social media: "I hate to say it, but we might actually be screwed."

The reason Seedance 2.0 is so powerful is that it adopts a multimodal input method.

Users can "feed" up to 12 assets to the system, including images, short videos, and audio files, as hard references for AI generation.

This means if you want to produce a video of a certain singer performing in a cyberpunk world, you only need to upload their photo and a piece of strong rhythmic music, then describe the atmosphere.

The system will automatically handle the logic of physical movement, making the character's expressions and actions fit perfectly with the rhythm of the music.

For ordinary users, this might just be a toy for making funny videos to give us a laugh.

But for professional content creators, this is undoubtedly a revolution in productivity.

Whether making vertical videos for TikTok or long horizontal videos for YouTube, Seedance 2.0 can complete them in an extremely short time at a very low cost.

It even supports scene extension, character replacement, and frame-by-frame fine-tuning after generation.

However, behind this technological carnival lies the deep fear of Hollywood giants.

Just days after the release of Seedance 2.0, Disney and Paramount Pictures took swift action, sending cease-and-desist letters to ByteDance.

The Motion Picture Association also issued a statement accusing ByteDance of unauthorized use of copyrighted works on a massive scale.

Their anger is not without reason, because these AI-generated videos are obviously trained based on protected content.

Those realistic motion captures of Jet Li likely come from his past movie footage.

And Jackie Chan's signature moves, Kanye's voice and image—these are all intellectual property protected by law.

Even trickier are the deepfake videos involving superstars like Tom Cruise and Brad Pitt.

These two actors themselves represent immense commercial value and film IPs, such as Cruise's "Mission: Impossible" series.

When AI generates scenes they never acted in, clothes they never wore, or even words they never said, this seriously infringes on their "right of publicity," which is the right to control the commercial use of their image.

The biggest point of controversy currently is that the outside world is not clear on exactly what data Seedance 2.0 used for training.

ByteDance has not yet fully disclosed the sources of its training datasets.

But considering the model can so perfectly replicate the demeanor and mannerisms of specific celebrities, studios have reason to speculate that the model has devoured massive amounts of movie clips, photos, and recordings.

Disney is particularly nervous about this, as they hold decades of classic IPs like Marvel and "Star Wars."

If anyone can use AI to generate high-quality videos of Marvel characters, this will directly threaten Disney's commercial foundation.

Although OpenAI had previously reached an agreement with Disney to allow the use of Disney characters in the Sora generator, that was, after all, built on a foundation of commercial cooperation.

Whereas Seedance 2.0's current state of "barbaric growth" clearly touches the industry's bottom line.

It is undeniable that technological advancements like Seedance 2.0 are devolving content creation rights to the general public in an unprecedented way.

Anyone with a laptop or even a mobile phone now has the ability to produce Hollywood-level videos.

No need for massive crews, no need for expensive budgets, and no traditional industry barriers.

This is absolutely a game-changing moment for independent filmmakers, short video creators, and small businesses.

But the other side of the coin is that the boundaries of licensing and ethics are being thoroughly shattered.

These stars replicated by AI never agreed to be used in this way.

This leads to a broader social question: Should AI be allowed to replicate human identity traits without permission?

From a legal perspective, studios are striving hard for a "no" answer.

But from a creative perspective, internet users are cheering, "Who cares, this is too cool."

From a technical perspective, ByteDance has not stopped its pace, and other tech companies are also racing to develop similar tools.

Once Pandora's box is opened, it is almost impossible to stuff the genie back in.

The future direction is likely to be a hybrid model: establishing strict celebrity licensing agreements, requiring creators to label use of public figures' images, and building a clear legal framework to define the scope of fair use for AI generation.

But at least for now, in this brief window before legal regulation fully lands, we are in a frantic and fascinating moment of chaos.

You can freely download those AI fighting videos of Jet Li, or forward clips of Kanye singing in Chinese.

Because this might be the only time we can freely enjoy this possibility of infinite creation before the legal gates close forever.