CALIFORNIA: Meta Platforms is betting that users will trust it with their secrets, provided it promises not to look. The company has launched Incognito Chat, a new privacy feature for WhatsApp and the Meta AI app, designed to create temporary AI sessions that even Meta itself cannot access. It is a bold claim from a company not historically celebrated for its restraint with user data.
The feature runs on Meta’s Private Processing infrastructure, which the company says creates isolated AI sessions processed in a protected environment without being stored by default. Once the session ends, the conversation vanishes automatically and leaves no trace in the user’s regular AI chat history. Meta insists this goes further than the standard incognito modes offered by rival AI services, arguing that the system’s architecture makes it structurally impossible for the company to view either user prompts or the AI’s responses.
The timing is deliberate. Users are increasingly taking their most sensitive queries — health anxieties, financial worries, relationship troubles — to AI chatbots. Meta is clearly aware that privacy concerns could deter uptake of its AI products, and Incognito Chat is its answer to that hesitation. A technical whitepaper outlining the security model has been published as the company attempts to substantiate its claims.
More is coming. Meta has also announced Side Chat with Meta AI, an upcoming feature that will allow users to access AI assistance during active WhatsApp conversations without switching screens. It will operate under the same Private Processing architecture as Incognito Chat. Both features will roll out gradually over the coming months.
Incognito Chat can be activated directly within WhatsApp or through the standalone Meta AI app, running as a separate session from standard AI interactions.
