The AI Takeover of YouTube Is About to Get Supercharged

Like all generative AI video models, Google's Veo 3 shines in small doses. At the company's Made On YouTube event this week in New York, journalists and creators saw just what it can do in YouTube Shorts. 

You could, for instance, film a video of the streets of New York and add a penguin riding a skateboard. Or you could have AI manipulate a picture of yourself to have a black belt in karate or dance like a ballerina. One demo showed how Veo 3 could insert a realistic-looking monster into a living room, with the model accurately casting shadows across the ceiling. 

It makes sense that Google would bring Veo 3 to its short-form video product, which is in direct competition with TikTok. 

And it's impressive technology. Giving creators this level of power will spark new and novel videos, ones that might have required teams of 3D effects artists to implement in the past. 

Still, it's hard not to be concerned. With AI proliferating on every corner of the internet with bogus articles, scams, spam and low-quality slop, YouTube might be unleashing a beast onto its unsuspecting 2.5 billion monthly active users. 

Don't miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.

"I'm praying that YouTube will add clear, automatic AI labels to any videos generated or altered with Veo," said Joss Fong, a video journalist and host of HowTown. "It would be a strong signal that Google understands [its] responsibility to help us navigate the surreal and confusing experience of consuming videos these days." 

At the very least, Fong trusts Google, as it created an industry-leading AI watermarking system, but she wants the company to signal it clearly to viewers. Even with watermarks and disclosures, people can still fall for AI content or, in Fong's case, begin questioning everything they see online, including videos of themselves. She's confident that YouTube will catch the most harmful misinformation, especially regarding world events or major public figures. 

When asked about bad actors using this tech to make potentially dangerous, misleading or harmful content, in a bid to grab views, YouTube's director of product management for shorts and GenAI creation, Dina Berrada, said it's a consequence of technical evolution.
"Well, first YouTube started out, people complained about 20 cat videos," Berrada said. 

YouTube then evolved to more long-form content and, with the popularity of platforms like TikTok, introduced Shorts. That, too, got complaints from users, according to Berrada. Now, YouTube Shorts accounts for 70 billion views per day. 

Berrada said it'll take time for creators to learn how to use Veo 3 to produce high-quality content. The company is also learning what kind of content should be recommended to viewers. "We're not looking only to optimize for pure engagement, but we're really looking at viewer satisfaction and valued watch time," Berrada said. 

YouTube's new AI features come as Google embeds the technology across seemingly every product in its portfolio, from Google Search to Chrome. Earlier this week, the search giant became just the fourth company to hit a $3 trillion market valuation. 

YouTube is the most popular streaming service in the world. With billions of active users, it's challenging to find fresh avenues for growth. New AI-powered auto-dubbing features are a powerful way to let English or Spanish videos find original audiences in Portugal or Italy. 

Fong, who started out making videos for Vox before spinning off and starting HowTown, has seen YouTube's growth in the wake of traditional media's contraction. 

"If a video performs well, that means people want to watch it," said Fong.

Traditional studios, showrunners, critics, stars and executives no longer dictate what ascends in popularity like they did decades prior. Now, it's directly people's likes, plus algorithmic thrust, which dictate popularity in the age of the internet.  "With the gatekeepers and tastemakers now deceased, we get the media we deserve."

According to Paul Erickson, principal analyst in the media entertainment practice at Omdia, giving creators these tools acts as a "force multiplier," which then helps proliferate YouTube's short-form content. "That then leads, by extension, to more traffic, more consumption, and, further down the road, more ad revenue."