Eli Collins, vice president of product management at Google DeepMind, first demonstrated generative AI video tools to the company’s board of directors in 2022. Despite the model’s slow speed, high operating cost, and sometimes inaccurate results, he says it was an eye-opening moment for them to see new video clips generated from a random prompt.
Now, just a few years later, Google has announced plans for a tool within the YouTube app that will allow anyone to generate AI video clips, using The Veo model of the companyand publish them directly as part of YouTube Shorts. “Starting in 2025, we’re going to enable users to create standalone video clips and shorts,” says Sarah Ali, senior director of product management at YouTube. “They’ll be able to generate six-second videos from an open-ended text prompt.” Ali says the update could help creators looking for material to fill out a video or trying to imagine something fantastical. She’s adamant that the Veo AI tool isn’t meant to replace creativity, but rather augment it.
This isn’t the first time Google has introduced generative tools for YouTube, though this announcement will be the company’s most extensive AI video integration to date. Over the summer, Google launched an experimental tool, called Sleep Screento generate AI backgrounds for videos. Before the full release of generated clips next year, Google will update that AI green screen tool with the Veo model sometime in the next few months.
The growing tech company has showcased multiple AI-powered video models in recent years, such as Image and LumiereBut it’s trying to coalesce around a more unified vision with the Veo model. “Veo will be our model, by the way, going forward,” Collins says. “You shouldn’t expect five more models from us.” Yes, Google will likely launch another video model in the future, but it expects to focus on Veo for the foreseeable future.
Google is facing competition from several startups developing their own generative text-to-video tools. OpenAI’s Sora is the best-known competitor, but the AI video model, announced in early 2024, is not yet publicly available and is reserved for a small number of testers. As for tools that are widely available, AI startup Runway has released several versions of its video software, including a recent tool to adapt original videos into alternate reality versions of the clip.
YouTube’s announcement comes at a time when generative AI tools have become even more controversial for creators, who sometimes view the current wave of AI as stealing from your job and trying to undermine the creative processAli doesn’t believe generative AI tools will come between creators and the authenticity of their relationship with viewers. “It’s really about the audience and what they’re interested in, not necessarily the tools,” he says. “But, if your audience is interested in how you made it, that will be open through the description.” Google plans to watermark each AI-generated video for YouTube Shorts with Synthesis identificationwhich incorporates an invisible label to help identify the video as synthetic, as well as including an “AI-powered” disclaimer in the description.
Hustle culture influencers are already trying play the algorithm by using multiple third-party tools to automate the creative process and make money with minimal effort. Will next year’s Veo integration lead to a new flood of low-quality, spammy YouTube Shorts dominating users’ feeds? “I think our experience with recommending the right content to the right viewer works in this world of AI at scale, because we’ve been doing it at this large scale,” Ali says. He also notes that Standard YouTube Guidelines It remains valid no matter what tool is used to create the video.
AI art often has a distinctive aestheticswhich could be concerning for video creators who value individuality and want their content to feel unique. Collins hopes that Google’s fingerprints aren’t all over AI video results. “I don’t want people to see this and say, ‘Oh, that’s DeepMind’s model,’” she says. Getting the cue to produce an AI result aligned with what the creator envisioned is a central goal, and avoiding obvious aesthetics for Veo is critical to achieving wide-ranging adaptability.
“A big part of the process is building something that is useful for people, scalable and deployable,” Collins says. “It’s not just a demo. It’s being used in a real product.” Collins believes putting generative AI tools inside the YouTube app will be transformative for creators, as well as for DeepMind. “We’ve never built a product for creators,” he says. “And we’ve certainly never done it at this scale.”