AI-FRONTIERMay 13, 20261 min read11 views

Mark Zuckerberg announces ‘completely private’ encrypted Meta AI chat

Meta CEO Mark Zuckerberg says its new Incognito Chat is “the first major AI product where there is no log of your conversations stored on servers.”

Meta CEO Mark Zuckerberg says its new Incognito Chat is “the first major AI product where there is no log of your conversations stored on servers.” Messages in Incognito Chat aren’t saved or stored in users’ chat history, similar to incognito modes on other AI chatbots, but Meta says its version is different because it also uses end-to-end encryption, which Meta recently removed from Instagram DMs:

Mark Zuckerberg announces ‘completely private’ encrypted Meta AI chat

Incognito Chat AI messages disappear after users leave their chat session, which Meta says makes it different from other bots.

“Other apps have introduced incognito-style modes, but they can still see the questions coming in and the answers going out. Incognito Chat with Meta AI is truly private, meaning no one — not even Meta — can read your conversations.”

Google says it keeps data from temporary chats in Gemini for up to 72 hours. Temporary chats in ChatGPT can be stored for up to 30 days, while incognito chats in Claude are kept for a minimum of 30 days. Anthropic and OpenAI did not immediately respond to a request for comment.

Logs from ChatGPT are central to recent lawsuits over mass shootings in Tumbler Ridge, Canada, and at Florida State University last year, and a lawsuit from the New York Times has included a court order to store conversations “indefinitely.” Google is being similarly sued by the family of a 36-year-old man who Gemini allegedly instructed to carry out a series of “missions” leading up to his death.

Meta’s Incognito Chat is built on the same Private Processing tech launched last year for processing data in WhatsApp and will roll out “over the coming months” in both WhatsApp and the Meta AI app.

Follow topics and authorsfrom this story to see more like this in your personalized homepage feed and to receive email updates.

Source: The Verge AI

More in this category

ai-frontier

Here are the 30,000 songs Sony is suing Udio’s AI music generator over

Sony Music has filed a new lawsuit against AI music generator Udio, alleging copyright infringement of over 30,000 songs after a judge blocked them from being added to an existing lawsuit.

ai-frontier

SpaceX in your index fund, explained

Index funds are touted as one of the safest ways to invest. But does the fast-tracked inclusion of a highly volatile giant like SpaceX threaten their stability?

ai-frontier

China’s AI models have Trump’s AI world at war with itself

The launch of Moonshot's free, open-source Kimi model has triggered a fierce debate among Donald Trump's AI advisors, exposing deep divisions over how to protect US tech dominance against cheaper Chinese rivals.

ai-frontier

Adobe’s ‘natural look’ camera app embraces generative AI

Adobe's experimental camera app, Project Indigo, which was originally designed to bring a natural SLR-like look to iPhone photography, is receiving a major update integrating generative AI tools powered by Google's AI model instead of Adobe's own Firefly.

ai-frontier

China delivers a one-two punch to America’s AI dominance

China’s leading AI companies are ramping up the pressure on Silicon Valley, as Moonshot and Alibaba unveiled models they claim can go toe-to-toe with the best from OpenAI and Anthropic at a fraction of the cost.

ai-frontier

Safety and alignment in an era of long-horizon models

Long-running models can solve difficult, open-ended problems, but their persistence gives them more opportunities to take unwanted actions. This article explores how OpenAI addressed novel safety failures in these models through trajectory-level monitoring and improved alignment.