How soon will it be before Netflix subscribers begin to see movies and TV show that were made with generative AI technology? According to CEO Ted Sarandos, it’s already happened.

In a video conference call after Netflix’s earnings were announced on July 17, Sarandos took questions, one of which was focused on when and how the streaming company will generate content with artificial intelligence tools.

Surprisingly, Sarandos said the company already did it, on an Argentinian sci-fi show called El Eternauta, a graphic-novel adaptation known as The Eternaut in English-speaking markets. For a scene showing a building collapsing in Buenos Aires, Netflix’s tech team worked with the filmmakers to generate AI footage that was used as final footage.

Sarandos revealed that this was the first time that Netflix has used generative AI for final footage in any TV show or movie it has produced.

“We remain convinced that AI represents an incredible opportunity to help creators make films and series better, not just cheaper,” Sarandos said in the call. “Real people doing real work with better tools,” he added.

The Netflix chief said that the show’s budget would never have allowed footage like that to be created with traditional visual effects tools and workflows, and that it was produced 10 times faster than it would have been traditionally. “They were able to achieve an amazing result with remarkable speed,” Sarandos said.

Filmmakers, he said, are already using AI for previsualization and shot planning as well as visual effects such as de-aging. The company also plans to keep expanding its use of AI for keep improving recommendations and other features offered to subscribers.

El Eternauta has already been renewed for a second season and received positive reviews.

A streaming-industry shift

The surprise reveal from Netflix is in line with the company’s frequent early adoption of new technologies.

“Netflix using AI for final footage in El Eternauta isn’t entirely surprising given their track record of embracing new technologies early — from streaming to algorithm-driven recommendations to interactive content,” Kate O’Neill, founder of KO Insights and author of What Matters Next, told CNET. 

O’Neill was also Netflix’s first content manager. 

“What’s notable is how matter-of-factly CEO Ted Sarandos revealed this, suggesting they view it as a natural evolution rather than a watershed moment,” O’Neill says.

O’Neill predicts that with the speed of AI development and the economic pressures that streaming-content services are under, the use of AI tools in this way could become mainstream as soon as 18 to 24 months from now. “The technology is moving faster than the legal and ethical frameworks needed to govern it,” she says.

Some of those ethical and legal questions include what data AI models are being trained on and if intellectual property is being infringed upon, not to mention the effect widespread AI may have on an industry of actors, craftspeople and digital-effects workers. 

“While AI can democratize access to sophisticated visual effects, we need transparency about how these tools are trained and whether they’re incorporating existing copyrighted material,” O’Neill says.