Introducing Incognito Chat
Meta is rolling out a new feature called 'Incognito Chat' within WhatsApp, designed to offer users a more private way to interact with its AI. This mode
allows for completely confidential conversations with Meta AI. Once a user initiates an incognito chat, their messages will not be stored and will automatically vanish upon closing the chat interface. Meta explicitly states that neither the user nor Meta itself will have access to the content of these private exchanges. This development comes as Meta navigates various legal challenges and public scrutiny surrounding its AI technologies. The intention behind this feature is to encourage users to feel more comfortable asking deeply personal questions, ranging from financial advice and health concerns to sensitive work-related matters, knowing their interactions are protected.
Privacy Features Explained
The 'Incognito Chat' functionality is activated through a dedicated icon within the Meta AI chat interface. Upon activation, the conversation becomes ephemeral; messages are deleted once the app is closed or the phone is locked. Furthermore, Meta AI will reset and lose all conversational context from the incognito session, ensuring that each private chat begins as a completely fresh interaction. This feature is being gradually deployed across WhatsApp and the standalone Meta AI application over the upcoming months. In parallel, Meta is also developing a 'Side Chat' feature, which will permit users to privately query Meta AI within a group discussion without alerting other members to their AI interactions. This aims to provide discreet assistance without disrupting group dynamics or revealing personal queries to the wider group.
Considering Meta AI's Safety
The introduction of enhanced privacy features for Meta AI is accompanied by questions regarding the AI's overall safety and reliability, particularly in light of past incidents. It's worth noting that Meta AI has previously exhibited unusual behavior, such as mistaking a chatbot for a real person and expressing a desire to meet. Additionally, Meta faces ongoing legal battles, including allegations of illegally pirating copyrighted material to train its Llama models. A prominent science publisher has accused Meta of scraping copyrighted research papers without consent. These incidents may influence a user's decision regarding the level of personal information they feel comfortable sharing with Meta AI, even within its new private chat modes. The ultimate choice of whether to trust Meta AI with sensitive queries rests entirely with the individual user, considering the company's track record.
