Newsroom
Highlights:
In 2025, YouTube Shorts will add a new AI tool, enabling users to create six-second videos from text prompts.
This tool will use Google's Veo AI model to develop the text-to-video feature.
Google will ensure transparency by marking generated clips with SynthID watermarks and labels to indicate AI generation.
Get smarter at marketing in just 5 minutes
Our 1x weekly, bite-sized newsletter will give you everything you need to know in the world of marketing:
YouTube has announced that it plans to introduce a generative AI tool. This feature will allow creators to produce six-second video clips from text prompts within the app. “And for the first time, you’ll also be able to generate 6-second standalone video clips for your Shorts,” the platform said.
Sarah Ali, YouTube’s Senior Director of Product Management, also confirmed the update. She explained, “Looking forward to 2025, we're going to let users create stand-alone video clips and shorts.”
How YouTube’s AI text-to-video works
The new AI tool will let users input text prompts. The Veo AI model will then turn them into short video clips. Ali noted that the tool aims to support, not replace, creativity. It lets users generate unique video content while preserving authenticity.
How YouTube integrates the text-to-video feature
The text-to-video feature will be integrated into YouTube using Google’s Veo AI video generation model, introduced in May. Unlike previous models, Veo combines Google’s past efforts into a single tool. Veo can create 1080p videos in various “cinematic and visual styles.”
The feature will also build on Google's past AI experiments. These include "Dream Screen," which creates AI-based video backgrounds. It was introduced last year.
Creators will access Veo through the “create” feature within YouTube Shorts. Dream Screen will generate four images from a prompt. A creator can then select one to convert into a six-second video.
How YouTube plans to maintain authenticity with the text-to-video tool
YouTube announced that it will watermark all AI-generated videos with SynthID technology. It will also label them for people to know they are AI-generated. This is to ensure transparency about the nature of the content and help viewers know the difference between human-made and AI-generated videos.
YouTube has updated its site to “inform viewers when the content they’re seeing is synthetic.” It said that “content created by YouTube’s generative AI products and features will be clearly labeled as altered or synthetic.”
Competing AI text-to-video models
YouTube is not the only platform with this feature. Companies like OpenAI and Runway have similar generative text-to-video technologies. OpenAI's Sora model, though not yet widely available, is a rival. Runway has already launched multiple versions of its video software.
09/19/2024
📰
Stories like this, in your inbox every Wednesday
Our 1x weekly, bite-sized newsletter will give you everything you need to know in the world of marketing: