OpenAI Launches SORA 2 Video Generator, Hollywood on Edge

OpenAI has unveiled its new AI-powered video generator, SORA 2, stirring quite a bit of anxiety in the entertainment industry. Many in Hollywood are making “nightmare” forecasts about the future of showbiz amidst this technological advancement.

SORA 2 enables users to create high-quality, movie-like videos simply by inputting descriptions. What has Hollywood particularly rattled is the program’s ability to generate realistic versions of copyrighted characters and familiar actors in a matter of minutes—something that traditionally takes months or years to achieve.

For instance, SORA 2 can produce videos that place characters like Pikachu into iconic films, such as Saving Private Ryan.

While the technology is impressive, some critiques have emerged. The program occasionally stumbles with dialogue, and some users report glitches that can distort the final product—glitches that generally vanish when the video plays at normal speed.

Despite its flaws, the capabilities of SORA 2 have left many in Hollywood uneasy. With just a voice command, users can create cinematic scenes without needing a studio, producers, actors, or other resources.

SORA 2 can tackle tasks previously thought nearly impossible for video generation, such as effectively emulating complex physical movements and scenarios.

The rapid evolution of this technology has sparked a debate over its legality. SORA 2 doesn’t generate entirely new videos; instead, it manipulates existing images, sounds, and videos from a large database, crafting new compositions based on user prompts.

For example, if someone wants to see Scarlett Johansson in a lightsaber duel with Mickey Mouse, SORA 2 can deliver that. Yet, legal questions loom: Can the program legally use Johansson’s image? What about depicting a scene from Star Wars? Are these uses permissible under intellectual property laws?

Social media conversations suggest that as long as people aren’t profiting from these creations, they should be allowed to “fair use” the content. However, rights holders contend that unauthorized use of their properties isn’t acceptable.

Currently, there are no established legal guidelines to address these issues, which may lead to a chaotic environment until laws catch up with technological advancements.

In response to these developments, rights holders are seeking ways to exclude their copyrighted materials from the datasets used to train SORA 2. It’s a complicated legal landscape.

As Richard Hogue, a business lawyer and podcaster, pointed out, “The law moves slowly, much slower than technology.” He speculated that OpenAI might ultimately be fine regarding its training datasets, as long as they aren’t sourced from pirated material.

For the moment, access to SORA 2 is restricted and available only through invitations, specifically for users in Canada and the U.S., with plans to extend to ChatGPT account holders in the future.

Concerns about the impact of SORA 2 on Hollywood are growing. An article in Slate has expressed a grim outlook, suggesting that such technologies could undermine traditional filmmaking and give anyone the ability to make films.

The article firmly states, “This is an existential moment for human-created entertainment. If artists and industry professionals can’t adapt now, the integrity of art could face dire consequences.”