Google Simplifies the Process of Creating Your Own Deepfake

YouTube Shorts is introducing an exciting new AI feature that allows creators to easily clone themselves on camera. This launch, teased earlier this year, highlights the platform’s complex relationship with AI content, as it balances the addition of innovative tools with the challenges of managing deepfake scams and impersonations.

With this new tool, users can create a digital avatar that mirrors their appearance and voice, which can be integrated into existing Shorts or used to craft entirely new videos. YouTube emphasizes that these avatars will “look and sound like you,” presenting them as a safer option for generating content with AI.

Creating an avatar involves a bit more than just a click. According to YouTube, users need to record a “live selfie” that captures their face and voice while following specific prompts. For optimal results, good lighting, a quiet environment, a background free of other people or faces, and holding the phone at eye level are recommended.

Once the avatars are ready, users can choose the “make a video with my avatar” option to generate clips based on prompts, which can last up to eight seconds. Additionally, users can add their avatars to “eligible Shorts” in their feed, although YouTube hasn’t clarified what qualifies a Short as eligible.

The AI avatar feature comes with strict guidelines. They can only be used in the creator’s original videos, and creators have control over whether their Shorts can be remixed. Users can delete their avatars or any videos featuring them at any time, and avatars that remain unused for three years will be automatically removed.

All videos featuring avatars will be clearly marked as AI-generated, complete with visible watermarks and digital labels like SynthID and C2PA, which are designed to authenticate AI-generated content.

Not everyone will have immediate access to this feature. YouTube has stated that the rollout will be gradual, without specifying a timeline or initial availability. Creators must be at least 18 years old and have an existing YouTube channel to use the tool.

This avatar feature adds to YouTube’s growing collection of AI tools for creators, which includes AI-generated video clips on Shorts, AI auto-dubbing, and a channel analytics chatbot. Many of these tools are powered by Google’s Gemini AI models, enabling users to transform photos into videos, create music, and generate realistic images from scratch.

The introduction of this feature comes as OpenAI, one of Google’s main competitors in the AI space, has decided to scale back its video generation efforts. The company recently announced the discontinuation of its Sora video tool after a year of challenges, including high costs and various copyright issues, making it a less appealing option for investors ahead of a planned IPO.