Meta announced on Wednesday the introduction of “incognito” conversations with its Meta AI chatbot on WhatsApp. These chats will be processed securely and remain private.
Users can initiate an incognito chat by selecting a new icon during one-on-one interactions with Meta AI. This feature will also be integrated into the standalone Meta AI app.
The rollout of incognito chats on WhatsApp and the Meta AI app will occur over the next few months.
According to Meta, incognito conversations aren’t saved and messages disappear once the chat is closed. These sessions will end if the app is closed or the phone is locked, resulting in Meta AI losing the chat context.
“People increasingly use AI for all kinds of private queries, from financial or health inquiries to advice on responding to messages,” Alice Newton-Rex, VP of Product at WhatsApp, mentioned in a call with TechCrunch. “We aim to provide the capability to ask such questions as discreetly as possible.”
Meta has been establishing the foundation for secure AI chats on WhatsApp. Last year, it introduced a private processing infrastructure to develop AI features without compromising end-to-end encryption. WhatsApp has since introduced features like AI-generated message summaries using this framework.
The new incognito chat is powered by Meta’s latest Muse Spark model, released last month, as noted by Newton-Rex.
Meta is also developing a feature called Side Chat, allowing users to privately query Meta AI within chats without alerting other participants.
At present, users must tag a message to query the AI assistant, visible to everyone in the chat. For private inquiries, text needs to be pasted in a separate chat.
ChatGPT and Claude also offer incognito options, while DuckDuckGo and Proton have introduced privacy-focused chatbots.
Meta’s push for private AI conversations comes as user interactions with AI chatbots are reportedly scrutinized legally, as noted by Reuters last month.