Meta unveils AI tool that can generate, edit videos based on text prompts

By Kurt Wagner

Facebook parent Meta Platforms Inc. debuted a new artificial intelligence tool that can generate or edit videos based on a simple text prompt, elevating competition with rivals like OpenAI and Google in the race to develop the world’s most advanced AI technology.
 

Meta’s product, Movie Gen, can create a new video up to 16 seconds long based on a text prompt. It can also use such prompts to generate audio for or edit an existing video, or even use a photo to create a customized video featuring a real person.
 

Movie Gen is only available for some internal employees and a handful of outside partners, including some filmmakers, though the social media company — which also owns Instagram, WhatsApp and Messenger — plans to build the product into its existing apps sometime next year.
 

 

Executives are still discussing how best to implement this integration, but the aim is for Movie Gen to encourage more people to create or edit video posts, said Connor Hayes, a Meta vice president focused on generative AI products. It will be “fun to use, helpful for creators, good for overall engagement in the apps, but we don’t have a concrete product plan of what it will look like at this point,” he said. 
 

The social networking leader is one of many major tech companies pushing into AI models focused on generating videos, which are more complicated and expensive to build than those that return AI-generated text. Microsoft Corp.-backed OpenAI has its own video generation tool, called Sora, that debuted earlier this year and can create videos up to a minute in length, though that technology isn’t available to the public yet. DeepMind, a subsidiary of Alphabet Inc.’s Google, also has a video generation tool, called Veo, which was unveiled earlier this year. 
 

Meta is waiting to roll out Movie Gen for several reasons, including the technology’s efficiency. It currently takes “tens of minutes” to generate a video based on a text prompt — too long for general consumers, who are likely to use this on their phone, Hayes said. 
 

But Meta is also “sorting out a bunch of really important problems around safety and responsibility,” Hayes said, including how to handle personalized videos so that a user can’t create an inappropriate or unflattering video featuring another person without their consent. “That’s going to be probably the most important problem to solve before we would make the personalization capability available to folks broadly.”This type of technology has been used in the past to create misleading so-called deepfakes of famous people, including US President Joe Biden and pop star Taylor Swift, though Meta executives have said they’re working on ways to “watermark” these creations so that people can tell they’re AI-generated.
 

Meta has made AI advancements a key priority for the entire company, and Chief Executive Officer Mark Zuckerberg has repeatedly talked about AI as a driver of user and revenue growth. In the short term, Zuckerberg has credited AI for helping improve the company’s content algorithms by showing people more relevant posts and advertisements. Eventually, Zuckerberg has said he believes AI will play an even larger role in powering its apps and other futuristic wearables Meta is producing, like smart glasses.
 

©2024 Bloomberg L.P.