Whatsapp might make an AI feature Google and OpenAI don't offer: an AI image of you
WhatsApp may soon introduce a remarkable AI-driven feature that sets it apart from competitors like Google and OpenAI—a personalized AI image of yourself. According to reports, WhatsApp is developing an image creator tool that utilizes artificial intelligence (AI) to create avatars of users. These avatars can then be placed into various virtual settings suggested by the users themselves.
The upcoming feature, discovered in a future beta version of WhatsApp for Android by WABetaInfo, is likely to leverage the AI models from Meta, WhatsApp's parent company. By uploading a few photographs, Meta's AI Llama model will generate an AI-powered avatar that can be inserted into images based on specific text prompts.
To employ this feature, users will instruct Meta AI to "imagine" them in the desired chat. For instance, typing "@Meta AI imagine me…" in a chat will elicit the AI's image generation capability. It's essential to note that the feature is currently in a limited beta testing phase with selected users, and it remains uncertain when it will be widely accessible.
WhatsApp has previously introduced AI-related features such as in-app custom stickers generated from text prompts. However, incorporating the ability to embed oneself into AI-generated images would be a significant advancement for the messaging platform.
This concept not only appears appealing for WhatsApp but also distinguishes the feature from its peers at Meta. Most AI image generators offered by companies like OpenAI and Google are typically hesitant to create AI avatars of real individuals, often declining the requests altogether. For example, ChatGPT commonly states its inability to create AI images of actual people. Similarly, Google Gemini, rather than replicating a person, is more focused on generating images of characters with similar likenesses or attire. Moreover, Google has recently facilitated the removal of unauthorized AI avatars on YouTube to address concerns regarding deepfakes.
Nevertheless, privacy and security remain significant concerns when generating AI avatars of real people. While WhatsApp has not fully disclosed the privacy measures involved with the AI-generated avatars, it is expected to implement strict rules upon the feature's release. The company will likely prioritize secure handling of user data and provide users with controls over the usage of their likenesses. Introducing the feature as optional demonstrates a step in the right direction. Considering that WhatsApp seems to be utilizing the Meta AI interface already employed across various Meta properties like Facebook and Instagram, it could serve as a testing ground for a broader future rollout.