WhatsApp Unveils 'Incognito Chat' for Private AI Interactions, Sparking Accountability Debate
By Tom Gray
•
Published: May 18, 2026
WhatsApp, the popular messaging platform owned by Meta, is introducing a new "incognito" mode for conversations with its integrated AI chatbot, Meta AI, designed to offer users unprecedented privacy. The move aims to allow individuals to engage with the artificial intelligence assistant on sensitive topics without fear of their discussions being stored or monitored.
The newly launched "Incognito Chat with Meta AI" feature ensures that conversations are entirely private and temporary. Messages exchanged within this mode are processed in a secure environment utilizing Meta's proprietary "Private Processing technology," meaning that even Meta itself cannot access the content. Crucially, these chat histories are not saved by default and automatically disappear once a user exits the session, preventing the information from being used to train Meta's future AI models. Will Cathcart, Head of WhatsApp, highlighted that the feature was developed in response to user demand for private AI interactions, particularly for queries concerning personal health, finances, or relationships, where users are reluctant to share intimate details with companies. Mark Zuckerberg, Meta's CEO, underscored the significance of this development, calling it the "first major AI product where there is no log of your conversations stored on servers".
While WhatsApp has long championed end-to-end encryption for human-to-human chats, the technology behind "Incognito Chat" differs, yet Meta asserts it offers an "equivalent" level of privacy protection. For now, the incognito mode supports only text-based interactions, with strict safety protocols in place to prevent the AI from responding to potentially harmful or illegal requests. The feature is rolling out to users on both WhatsApp and the Meta AI app over the coming months.
However, this heightened privacy also introduces a point of contention regarding accountability. Professor Alan Woodward, a cybersecurity expert at the University of Surrey, while acknowledging that the feature poses minimal risk to WhatsApp's broader security framework, expressed concerns about the potential "lack of accountability" should issues arise. The professor suggested that the disappearing nature of these chats, which cannot be retrieved by either the user or Meta, could hinder investigations into instances where AI responses might contribute to harm, misuse, or even tragic outcomes like suicide. He emphasized the significant trust users are placing in the AI not to mislead them. The inherent risk of AI "hallucinating" or generating inaccurate information, coupled with the absence of oversight in incognito mode, presents a genuine concern for reliability. This debate is not new, as other AI companies have previously faced lawsuits related to their AI systems.
This new feature builds on Meta's broader AI strategy, which saw the introduction of Meta AI into WhatsApp last year, a move that initially drew criticism from users who were frustrated by the inability to disable it. Looking ahead, Meta also plans to introduce "Side Chat with Meta AI," which will enable users to privately consult the AI within existing human conversations, also leveraging the secure Private Processing technology.
The introduction of "Incognito Chat" reflects a growing industry trend among AI chatbot providers, including Google's Gemini and OpenAI's ChatGPT, to offer users more control over their chat history and data usage. As AI becomes increasingly integrated into daily communication, balancing user privacy with the need for safety and accountability remains a complex challenge for technology companies.
The newly launched "Incognito Chat with Meta AI" feature ensures that conversations are entirely private and temporary. Messages exchanged within this mode are processed in a secure environment utilizing Meta's proprietary "Private Processing technology," meaning that even Meta itself cannot access the content. Crucially, these chat histories are not saved by default and automatically disappear once a user exits the session, preventing the information from being used to train Meta's future AI models. Will Cathcart, Head of WhatsApp, highlighted that the feature was developed in response to user demand for private AI interactions, particularly for queries concerning personal health, finances, or relationships, where users are reluctant to share intimate details with companies. Mark Zuckerberg, Meta's CEO, underscored the significance of this development, calling it the "first major AI product where there is no log of your conversations stored on servers".
While WhatsApp has long championed end-to-end encryption for human-to-human chats, the technology behind "Incognito Chat" differs, yet Meta asserts it offers an "equivalent" level of privacy protection. For now, the incognito mode supports only text-based interactions, with strict safety protocols in place to prevent the AI from responding to potentially harmful or illegal requests. The feature is rolling out to users on both WhatsApp and the Meta AI app over the coming months.
However, this heightened privacy also introduces a point of contention regarding accountability. Professor Alan Woodward, a cybersecurity expert at the University of Surrey, while acknowledging that the feature poses minimal risk to WhatsApp's broader security framework, expressed concerns about the potential "lack of accountability" should issues arise. The professor suggested that the disappearing nature of these chats, which cannot be retrieved by either the user or Meta, could hinder investigations into instances where AI responses might contribute to harm, misuse, or even tragic outcomes like suicide. He emphasized the significant trust users are placing in the AI not to mislead them. The inherent risk of AI "hallucinating" or generating inaccurate information, coupled with the absence of oversight in incognito mode, presents a genuine concern for reliability. This debate is not new, as other AI companies have previously faced lawsuits related to their AI systems.
This new feature builds on Meta's broader AI strategy, which saw the introduction of Meta AI into WhatsApp last year, a move that initially drew criticism from users who were frustrated by the inability to disable it. Looking ahead, Meta also plans to introduce "Side Chat with Meta AI," which will enable users to privately consult the AI within existing human conversations, also leveraging the secure Private Processing technology.
The introduction of "Incognito Chat" reflects a growing industry trend among AI chatbot providers, including Google's Gemini and OpenAI's ChatGPT, to offer users more control over their chat history and data usage. As AI becomes increasingly integrated into daily communication, balancing user privacy with the need for safety and accountability remains a complex challenge for technology companies.