Meta launches private AI chat mode on WhatsApp with claim even it cannot read conversations
By Tanveer Ahmed :
Meta has launched a new “Incognito Chat” feature for its AI assistant on WhatsApp, introducing what the company describes as a fully private AI conversation mode that even Meta itself cannot access.
The feature, now rolling out on WhatsApp and the Meta AI application, marks one of the strongest privacy commitments the company has made regarding artificial intelligence services and comes amid growing global concerns over how AI firms collect, process and store user data.
According to Meta, Incognito Chat is built on a system called “Private Processing”, designed to ensure conversations with the AI assistant are processed inside a secure and isolated environment where the content remains inaccessible even to Meta’s own infrastructure.
The company says messages in Incognito Chat are not stored, disappear by default and cannot be viewed by Meta, WhatsApp or other users. Meta claims the technology extends the privacy protections associated with WhatsApp’s end-to-end encryption to conversations between users and artificial intelligence systems.
Industry analysts say the announcement represents a notable shift in how large technology companies are approaching AI privacy. While many AI platforms offer “private” or “incognito” modes, experts note that user prompts and responses are often still visible to the company operating the system.
Meta argues its approach differs because the data remains encrypted and inaccessible during processing, preventing the company from reading questions or using them for advertising or AI training purposes.
The company has published a technical whitepaper outlining the architecture behind the feature, stating that the processing occurs in a segregated computing environment designed to block even system operators from viewing conversations in plain text.
The launch reflects the increasing sensitivity surrounding the kinds of questions users ask AI systems. Analysts say many people now use chatbots for advice and information on highly personal topics including health, finance, relationships, work issues and mental wellbeing.
Privacy advocates have long warned that AI conversations can reveal deeply personal information that may be more sensitive than ordinary social media or messaging activity. Questions about medical symptoms, legal concerns or emotional struggles can create highly detailed profiles of users if stored or analysed by technology companies.
Meta says the new system is intended to address those concerns by ensuring no readable data trail exists within the company’s systems.
The company also announced plans for another feature called “Side Chat”, expected to launch in the coming months. The tool would allow users to privately consult Meta AI while participating in existing chats or group conversations without alerting other participants.
According to Meta, Side Chat would provide contextual AI assistance linked to ongoing conversations while preserving privacy protections under the same Private Processing system.
However, the launch has also raised questions among digital safety experts about how AI harms and crisis situations could be managed in an environment inaccessible even to the platform operator.
Concerns surrounding AI chatbots have intensified following several lawsuits and reports involving users who allegedly developed harmful emotional dependence on AI systems or received unsafe responses during mental health crises.
Critics argue that while stronger privacy protections are beneficial, they may also remove opportunities for platforms to detect dangerous interactions, including signs of self-harm or suicidal ideation.
Meta has not publicly explained how safeguarding measures or crisis intervention systems would function inside a processing environment it claims cannot be monitored.
The rollout of Incognito Chat comes shortly after Meta faced criticism over changes to privacy features on Instagram, where the company reportedly scaled back certain encrypted messaging protections.
Technology analysts say the success of Meta’s new AI privacy model could influence how other companies approach secure AI systems in the future, particularly as governments and regulators increase scrutiny of artificial intelligence, data collection and online safety.
About Author
