AI privacy

WhatsApp adds private AI chats that Meta says it cannot read

The new “incognito” mode is aimed at sensitive chatbot conversations, but a cyber security expert warns disappearing records could make harm harder to investigate

Source language: English
0
WhatsApp adds private AI chats that Meta says it cannot read
Location
Whatsapp
Avenue Lumumba, Mbamu, Kinshasa, Democratic Republic of the Congo
WhatsApp has introduced private Meta AI chats, saying conversations will not be stored in server logs, as experts raise accountability concerns.
Artificial intelligence Cybersecurity Meta AI Privacy WhatsApp

WhatsApp has introduced private Meta AI chats, saying conversations will not be stored in server logs, as experts raise accountability concerns.

WhatsApp has introduced an “incognito” mode for conversations with its Meta AI chatbot, offering exchanges that company leaders say cannot be read by Meta and will not be stored as chat logs on its servers.

The feature is designed for users who want to ask an AI assistant about sensitive subjects without leaving an accessible record. Will Cathcart, the head of WhatsApp, said people want answers on topics such as health, relationships and finances but may feel uneasy sharing personal information with the company.

When the mode is activated, neither the user’s messages nor the chatbot’s responses are monitored, and previous exchanges disappear from the user’s chat, according to the announcement reported by the BBC. Meta chief executive Mark Zuckerberg described it as the “first major AI product where there is no log of your conversations stored on servers.”

The privacy move also raises a harder question: what happens if an AI response causes harm and no one can retrieve the conversation afterward?

Prof Alan Woodward, a cyber security expert at Surrey University, told the BBC that the new system carries a low risk of weakening WhatsApp’s existing security. But he warned that disappearing conversations could limit accountability if an AI chatbot gives dangerous or misleading advice.

“Personally I think what you ask an AI should remain private as some people ask it very personal matters - but you are placing a great deal of trust in the AI not to lead users astray,” Woodward said.

The concern is especially acute because AI companies have faced lawsuits over alleged harms linked to chatbot interactions. If neither users nor Meta can recover a chat history, it may be difficult to establish what was said before an incident involving harm, death or suicide.

Cathcart said the incognito feature will initially handle text rather than images, and that Meta AI’s guardrails will be cautious about refusing requests that could be interpreted as harmful or illegal. He also said the technology behind the new mode is not the same as WhatsApp’s end-to-end encryption for ordinary messages, but called it the equivalent for this purpose.

The launch comes as Meta pushes AI deeper into its apps. Meta AI was added to WhatsApp last year and drew criticism from some users who objected to being unable to turn it off. In May 2025, Zuckerberg said Meta AI had reached one billion users across Meta’s apps, which include WhatsApp, Instagram, Facebook and Messenger.

WhatsApp has blocked other AI chatbots from operating through its systems, meaning Meta’s own assistant is the only chatbot available to its billions of users on the platform. The unanswered test for the new private mode is whether Meta can preserve the confidentiality it is promising while still providing enough oversight when something goes wrong.

More from this section

Tech news

Figures mentioned

More from this location

Related tags

Related articles

Shared tag: Artificial intelligence OpenAI trial
Altman defends his leadership as Musk lawsuit tests OpenAI’s mission

The OpenAI chief told a federal jury in Oakland that he is trustworthy and said Musk sought lasting control of the ChatGPT maker before leaving the company

May 13, 2026 Oakland
Shared tag: Artificial intelligence U.S.-China tech race
China Was Denied Access to Anthropic’s Newest AI, Report Says

The limited source summary says newer Anthropic and OpenAI models are widening the U.S. edge over China, but key details remain unavailable

May 12, 2026 anthropic features
Shared tag: Artificial intelligence OpenAI trial
Altman says Musk wanted control of OpenAI, even after death

The OpenAI chief told a federal jury in Oakland that Elon Musk repeatedly sought control of the company and supported a for-profit shift he now challenges in court

May 12, 2026 Oakland
Shared tag: Artificial intelligence Energy and AI
AI Data Center Boom Puts New Pressure on Household Power Bills

In Georgia, CBS News found six Georgia Power rate hikes in three years as advocates warn that surging electricity demand from data centers could leave residents exposed

Apr 28, 2026 Atlanta
Shared tag: Artificial intelligence Tech jobs
Meta to cut 8,000 jobs as Microsoft readies U.S. buyouts

Meta says the layoffs will begin May 20, while Microsoft is preparing voluntary buyout offers for thousands of U.S. employees as AI costs rise across the sector

Apr 27, 2026 Meta
Shared tag: Artificial intelligence Artificial intelligence
Amp Raises $1.3 Billion for an A.I. ‘Grid

The start-up is seeking to build an alternative in a market where major technology companies control much of the hardware needed for advanced A.I

May 12, 2026

Comments (0)

Please log in to comment.
No comments yet.