- YouTube launches AI avatars for Shorts that mimic the creator's face and voice.
- The feature requires a "live selfie", is only for those over 18, and is being rolled out gradually.
- The generated clips last up to 8 seconds and are labeled as AI content with authenticity marks.
- Creators retain control to delete their avatar and generated videos whenever they wish.
YouTube has begun rolling out a new AI feature in Shorts that allows creators to create digital avatars of themselves . With this tool, any user with an active channel and over 18 years of age can generate a virtual version of themselves to appear in short videos without having to be on camera each time.
This new feature comes at a time when short videos account for a large portion of online consumption and when warnings are growing about deepfakes, impersonation , and the flood of low-quality, synthetic content, known in the industry as " slop ." YouTube is trying to strike a delicate balance: integrating generative AI into its ecosystem to facilitate content production, but with more visible security and transparency measures to prevent abuse.
What are the new AI avatars for YouTube Shorts?
The platform describes these avatars as digital representations that “will look and sound like you .” In other words, the system replicates the creator's face, main gestures, and voice from a guided recording, and uses that model to generate Shorts of up to about eight seconds in length per clip.
In practice, they function as a kind of "virtual double" that can star in videos without the creator having to record themselves each time . The user can write a description or a short script and let the AI produce the clip with their avatar, or insert it into already published Shorts that meet certain technical criteria.
The strategy relies on YouTube's creation environment and Google's Gemini models , which are already used for features like automatic clips, AI dubbing, and image generation. Avatars are not conceived as an isolated element, but rather as one piece of a broader strategy to integrate generative tools within YouTube itself.
For many creators, this feature can serve as a middle ground: maintaining their personal presence and style without constantly being on camera . This can be especially useful for those who are shy, want to partially protect their image , or simply want to streamline their daily workflow.
Requirements and availability: who can use the avatars
YouTube has opted for a phased rollout. The company indicates that the feature will be activated gradually worldwide , without a fixed public schedule or a detailed list of priority countries, which is typical for this type of controlled testing.
What is clear are the basic requirements: only users over 18 years of age with an existing YouTube channel will be able to use avatars . The age restriction aims to limit the use of this technology by minors, given that it is a very sensitive tool when working with real faces and voices.
The option is integrated into the YouTube mobile app and YouTube Create , the editing and production environment designed for creators. It hasn't yet been specified whether there will be significant differences between regions like the European Union and other markets, although the European regulatory framework for AI and data protection will likely influence the rollout and its timing.
For now, the feature is limited to Shorts format content . The platform has left the door ajar for bringing this technology to longer videos in the future, but that step will depend on the reception, the results of initial tests, and the balance between usefulness and perceived risks.
How to create your digital avatar step by step
The creation process is designed to be relatively simple, although it requires following a series of instructions to ensure the model is of sufficient quality. From the YouTube mobile app, the general process is as follows, with slight variations depending on the version:
- Open the YouTube app and access the “Create” option.
- Click on the Gemini icon or on the AI tools section located in the upper right corner.
- Within the creation section, select “Create a video with my avatar” or “Make a video with my avatar”.
- Follow the on-screen instructions to record a "live selfie" with your face and a voice sample.
- Wait for the system to process the information and generate the first version of the avatar.
During recording, YouTube recommends using good lighting , a quiet environment, and a background free of other people or visible faces . It also suggests holding the phone at eye level and moving only as much as necessary so that the AI can accurately capture facial features and voice.
Once the capture is complete, the platform displays a preview of the avatar . If the result is not satisfactory, the user can tap "Repeat" to record the samples again; if the model is satisfactory, it is confirmed with "Next" and saved to the account.
From that point on, the avatar becomes available as another resource in the creator's workflow. It's not necessary to repeat the process for each video, unless you want to refine the model or change its appearance for any reason.
Use the avatar in new Shorts and in videos already published
With the avatar stored, creators can use it in various ways within the Shorts environment. The most straightforward option is to generate new videos from scratch using the avatar as the main character.
To do this, simply go back to the “Create” section and choose the option to “make a video with my avatar” . The user writes a short description or script, and the tool produces a clip of up to about 8 seconds, which can then be combined with other fragments to create longer pieces or integrated into an existing montage.
Another possibility is to use avatars from previously published or created Shorts , provided those videos are deemed "eligible" by the platform. In this case, the general process is:
- Enter the Shorts feed and choose the desired video.
- Click on the “Remix” option and then on “Reimagine” (or equivalent depending on the interface).
- Select “Add me to this scene” to insert the avatar into the clip.
This approach allows, for example, updating old content, reacting to other videos, or appearing in collaborations without having to record new takes each time. Production becomes more flexible and, above all, faster for those who publish frequently.
In any case, YouTube emphasizes that the avatar can only be used by its creator on their channel and in content under their direct control. It is not a tool for creating third-party clones or for others to manipulate the model without permission.
Avatar control, deletion, and usage limits
One of the key features of this function is the ability to manage the avatar afterward. YouTube emphasizes that the creator retains full control over their avatar from their account. This includes repeating the creation process, updating it, or deleting it entirely.
By deleting an avatar, the platform agrees not to retain the materials used during the generation process . Furthermore, it has established an expiration policy: if an avatar remains inactive for three years, it will be automatically deleted from the systems.
It's important to note that deleting an avatar doesn't automatically delete the Shorts it appears in . If the creator wants those clips removed from the channel, they'll have to delete them one by one, just like any other published video.
The tool also offers settings on how Shorts featuring the avatar can be remixed or reused . The channel owner can restrict others from editing or integrating these videos using remix functions, thus strengthening control over their digital image.
This set of limits does not resolve all possible risks, but it does make a difference compared to other digital cloning systems where the conditions for data erasure and retention are more opaque or depend on general clauses that are difficult to interpret.
AI labels, watermarks, and the fight against deepfakes
The launch of avatars coincides with a time of intense pressure on major platforms to clearly identify content generated or manipulated using AI . Both regulators and users are demanding greater transparency to prevent deception and malicious use.
In response, YouTube will integrate visible markers into all avatar-generated videos indicating that they are AI-produced content . These markers will appear in the player itself and in the description, so that any viewer can immediately recognize that they are watching a synthetic version of the creator.
In addition, the company will apply digital tags such as SynthID and C2PA , two standards designed to embed information about the origin and authenticity of multimedia content. These tags are not visible to the naked eye, but can be read by automated systems and other platforms to verify whether a video has been generated or modified using AI.
The introduction of these signals occurs in a context where, according to reports such as Kapwing's, low-quality AI-produced content is gaining traction , capturing advertising revenue and distorting the perception of reality in many people.
With Shorts avatars, YouTube is trying to demonstrate that it's possible to promote generative AI without further opening the door to malicious deepfakes . The actual effectiveness of these measures will depend, however, on how they are implemented in practice and whether other platforms, media outlets, and users take these labels seriously when sharing or validating videos.
Productivity for creators and doubts about authenticity
From the perspective of those who produce content daily, avatars can become a powerful tool for saving time and publishing more frequently . A single digital model can be reused across multiple pieces, generate quick responses to trends, and keep a channel active even when the creator can't physically record.
This logic of “automated production” aligns with the evolution of short videos, where consistency is crucial and audience attention spans are very limited . Being able to create several 8-second clips in quick succession from brief texts or ideas can make the difference between staying relevant and getting buried in the feed.
However, this same tool opens a debate about the extent to which it dilutes the authenticity and connection that many viewers value in creators. If the face and voice are synthetic, even if based on the real person, some users may perceive a certain "distance" or lack of spontaneity.
At the same time, the presence of digital clones raises questions about who truly controls the creator's image and identity in the long term . Even with deletion options and usage limits, the expansion of these technologies multiplies the scenarios in which an altered version of someone can circulate beyond their direct control.
In Europe, these concerns intersect with an increasingly stringent regulatory framework regarding generative AI, data protection, and online transparency. While YouTube has not publicly detailed all the specific adaptations for the European market, it is expected that regulatory compliance will play a significant role in how the feature is deployed and monitored in the region.
While questions remain about the timeline and specifics of its implementation in Europe, YouTube's move clearly demonstrates a trend: content creation is moving towards a model where the creator's presence can be replicated by AI with increasing convincingness . If watermarks, authenticity labels, and usage limits are rigorously enforced, the feature could solidify its position as a useful tool for producing Shorts more flexibly. Otherwise, the debate surrounding digital identity, consent, and trust in what we see on screen will continue to gain momentum in an environment where distinguishing between the real person and their synthetic version becomes increasingly difficult.
