WhatsApp appears to be developing a new generative AI feature that allows users to create personalized avatars for use in any imaginable scenario. This upcoming feature, discovered by WABetaInfo in the latest Android version of WhatsApp Beta 2.24.14.7, seems to combine user-provided images, text prompts, and the AI model Llama by Meta to generate the avatars.

A screenshot shared by WABetaInfo shows that users can imagine themselves in "any scenario from the forest to outer space." These examples look quite typical of images generated by AI generators, especially if you have used applications like Lensa AI or Snapchat's "Dreams" selfie feature.

According to the information card screenshot of the upcoming Imagine AI user avatar feature in WhatsApp, users need to take a selfie once to create a personalized avatar. Meta AI will then use these photos to train itself and generate similar images of the user in any scenario. Users can generate their avatars by entering "Imagine me" followed by a scene description in the Meta AI chat or by typing "@Meta AI imagine me..." in other WhatsApp conversations.

Reportedly, this feature will be optional and will require manual activation in the WhatsApp settings. WABetaInfo also states that reference images can be deleted at any time through the Meta AI settings.

It is currently unclear when this new Imagine AI selfie feature will be officially released. WhatsApp is still rolling out support for its Meta AI chatbot to US users, as well as a more general real-time AI image generation feature. Therefore, it may take some time for the new AI avatar feature to go live. Given the issues previously encountered with Meta's generative AI tools, caution is advised.