WhatsApp is testing AI stickers made from your own face

Sending stickers in chat is second nature by now, but turning your own face into a custom sticker exploring outer space is a completely different ballgame.

According to findings by WABetaInfo, Meta is currently developing a new generative AI tool within WhatsApp that lets users create customized stickers using their actual likeness. The feature was spotted in the Android beta version 2.24.14.7.

Unlike traditional AI art generators that spit out generic fantasy portraits, this tool takes your real face and drops it directly into whatever prompt you throw at it.

Selfies, prompts, and a blue ring

Setting it up starts with a brief onboarding process. WhatsApp prompts you to take and upload a handful of reference selfies. Meta AI analyzes these facial cues to understand what you look like.

Once the setup is done, bringing yourself into the conversation is straightforward. In any chat window, typing "@Meta AI imagine me" alongside a description—like trekking through an enchanted forest or floating inside an astronaut suit—triggers the assistant to generate a personalized sticker with your likeness.

Privacy has always been a sensitive issue for Meta, especially when personal biometrics are involved. The company appears to be making the feature strictly opt-in. The setup photos remain under user control and can be deleted at any moment through the Meta AI settings menu.

For now, the sticker feature remains locked behind closed beta testing. Even Meta AI itself is still rolling out gradually in select markets like the US and India, signified by a distinctive blue icon on the main inbox screen.

Model switching and the battle for your keyboard

WABetaInfo also spotted another significant upgrade under the hood: the ability to switch between different AI models inside WhatsApp.

Users will reportedly be able to run routine queries using the default Llama 3-70B model, while toggling over to the flagship Llama 3-405B for heavier lifting. The catch is that Llama 3-405B will carry a weekly usage cap, which looks suspiciously like the groundwork for a future paid subscription tier similar to ChatGPT Plus.

Meta is far from alone in trying to turn your daily texts into an AI playground. Apple recently introduced Image Playground and Genmoji to turn personal photos into custom iMessage reactions, while Google is reportedly prepping a native Creative Assistant app for the Pixel 9 lineup to handle sticker creation.

Why it matters

For everyday users, this represents the ideal low-friction use case for generative image models. Most people do not care about crafting elaborate art prompts in Midjourney, but they love dropping absurd, hyper-personalized reaction images into group chats. Lowering that barrier to a single text prompt makes advanced tech immediately enjoyable.

For the tech landscape, this is about defending engagement. Apple controls the iOS keyboard with Genmoji, and Google controls Android with native Pixel tools. By piping Llama models directly into WhatsApp, Meta ensures its multi-platform messaging empire does not lose cultural relevance to OS-level AI gimmicks.

We think giving users a direct delete switch for their training photos was a necessary move to preempt privacy pushback. However, the real test will be content moderation. It looks inevitable that people will try feeding reference selfies of friends, ex-partners, or coworkers without consent to create mocking stickers. Until Meta proves it can effectively curb non-consensual face generation inside private chats, this clever parlor trick carries real harassment risks.