WhatsApp adds incognito AI

Messaging apps are where people ask the awkward questions. Health stuff. Money stuff. Relationship stuff. That is exactly why Meta’s latest AI move matters. On May 13, Meta launched Incognito Chat with Meta AI for WhatsApp and the standalone Meta AI app — a mode built for temporary conversations that the company says even Meta cannot read. ### What actually launched? Incognito Chat is a separate way to talk to Meta AI inside WhatsApp and the Meta AI app. The idea is simple — you open a private AI conversation, ask whatever you want, and that chat is treated as temporary from the start. Meta says the conversation is not saved and the messages disappear by default, so this is not just “AI in WhatsApp,” but “AI in WhatsApp without a lasting thread hanging around.” (about.fb.com) ### Why is WhatsApp the important venue? Because WhatsApp is already the place people use for their most private everyday conversations. Meta has been adding more AI across its apps for a while, but WhatsApp creates a problem normal AI products do not have — people expect the app to feel intimate and locked down. If AI is going to live there, it has to feel less like posting into a cloud service and more like whispering into a sealed room. (about.fb.com) ### How is this supposed to stay private? The technical hook is something Meta calls Private Processing. Meta described that system in April 2025 as a way to run AI tools for WhatsApp inside a secure environment designed so the company cannot access the contents being processed. Incognito Chat sits on top of that system. Basically, Meta is saying the privacy promise is not just a policy choice — it is built into how the request gets handled. (about.fb.com) ### Is this just “incognito mode” with new branding? Meta is very clearly trying to say no. The company’s launch post draws a distinction between incognito-style modes that still let the provider see prompts and outputs, and this new setup, which it says keeps the conversation invisible even to Meta. That is the whole pitch. Not merely fewer logs. Not merely hidden from your chat list. A system where the operator says it cannot inspect the exchange at all. (engineering.fb.com) ### Why launch this now? Because Meta is in the middle of turning AI into a layer across everything it owns. The Meta AI app launched in April 2025 as a standalone assistant connected to Meta’s broader ecosystem, and newer model rollouts have been positioned as feeding smarter AI across WhatsApp, Instagram, Facebook, Messenger, and glasses. Incognito Chat solves a very specific adoption problem in that push — people want AI help, but they do not want a permanent record of vulnerable questions. (about.fb.com) ### What does this change for users? It lowers the social and psychological cost of using AI. That sounds soft, but it is the whole game. A private AI mode makes it easier to ask embarrassing, high-stakes, or half-formed questions without feeling watched. If normal AI chat is like emailing a company help desk, this is closer to writing on a notepad that clears itself when you are done. That analogy is imperfect, but it gets at the emotional difference Meta is trying to create. (about.fb.com) ### What is the catch? The catch is that “private” here means private within Meta’s stated system design, not magic. Users still have to trust the implementation, the security model, and the boundaries of what counts as retained metadata. Meta’s public materials emphasize that message contents are protected and not saved in the chat, but privacy claims this strong will live or die on outside scrutiny and whether the system holds up under real-world testing. (about.fb.com) That last part is an inference from the architecture claims and the stakes involved. ### So what is the real story? Meta is not just adding another chatbot feature. It is trying to make AI feel safe enough to use in the most personal corner of its app empire. If that works, AI stops being a flashy extra and starts becoming a default part of private conversation. (about.fb.com)