Meta Platforms on Wednesday announced the launch of ‘Incognito Chat with Meta AI’, a new feature designed to allow users to have “completely private” conversations with its artificial intelligence chatbot on WhatsApp and the Meta AI app.
The feature, unveiled on May 13, 2026, is built on Meta’s “Private Processing technology” and runs all AI inference within a Trusted Execution Environment (TEE) — a secure infrastructure that Meta claims makes conversations inaccessible to external parties, including the company itself.
According to Meta, Incognito Chat offers three main privacy guarantees: exclusive user visibility protected by encryption, automatic deletion of conversations when the user exits, and private web searches that are not linked to the user.
“Once Incognito Chat is enabled, user chats with Meta AI become private and temporary. User prompts and AI-generated responses are not stored and will disappear by default,” the company said in its announcement.
Meta added that this is the “first major AI product where conversation logs are not stored on company servers.”
Rollout and access
The feature will begin rolling out on WhatsApp and the Meta AI app “over the coming months,” Meta confirmed. A new ‘Side Chat’ option will also arrive on WhatsApp, allowing users to privately access Meta AI inside chats with other users without disrupting the main conversation.
The company said Incognito Chat includes built-in safety guardrails. If users ask about harmful topics, the AI will steer toward helpful information, and may refuse to answer or end the session if the behavior continues. Users must be 13 or older to access the feature.
Context: Privacy debate intensifies
The announcement comes as AI-powered chat services face growing scrutiny over data privacy. Meta noted that users often ask chatbots sensitive questions involving financial, personal, health or work data. Will Cathcart, head of WhatsApp, said the feature was developed because “it doesn’t always feel like you should have to share the information behind those questions with the companies that run those AI systems.”
The launch also follows Meta’s decision on May 8, 2026 to discontinue the end-to-end encryption option for Instagram DMs, meaning Instagram can now access the content of direct messages, including images, videos and voice notes.
Meta’s blog post referenced “incognito-style modes” in other AI apps, stating that they don’t necessarily prevent platforms from accessing chat content. Rival services like Google’s Gemini and ChatGPT offer options to disable chat history or opt out of data training, but Meta is positioning Incognito Chat as a stronger guarantee that “no human beings, including anyone at Meta,” will view the contents.
A proposed class action filed last month against Perplexity alleged that its Incognito Mode quietly shares chat content with third parties, including Meta, for ad targeting.
The feature was announced by Meta CEO Mark Zuckerberg via Facebook and Threads, and by WhatsApp’s official accounts, using mockups that stated “Only you can see this chat” and “no one, not even WhatsApp or Meta, can ever see your conversations.”
Initial public reaction has been mixed. While some users called it a “game-changer for digital privacy,” others questioned whether the feature is truly private, with some commentators comparing it to “incognito mode on Chrome.”
