Substrate
ai

Meta Launches 'Incognito' Mode for WhatsApp AI Chats That Leaves No Trace or Access for Company or User

Meta Platforms is rolling out an incognito mode for WhatsApp that processes chats with Meta AI in a secure environment inaccessible to the company. Messages disappear by default when users exit a session and the feature requires age confirmation. The rollout comes as Meta AI reached a billion users last year and the company prepares to spend $145 billion on AI infrastructure in 2026.

Abc News
BBC News
TechCrunch
Wired
4 sources·May 13, 2:00 PM(1 hr ago)·2m read
|
Meta Launches 'Incognito' Mode for WhatsApp AI Chats That Leaves No Trace or Access for Company or UserAbc News
Audio version
Tap play to generate a narrated version.

Meta Platforms is introducing an incognito mode for WhatsApp that allows users to hold private conversations with its Meta AI chatbot. The mode processes messages in a secure environment that even Meta cannot access, the company said Wednesday. Incognito chat messages will not be saved by default and will disappear when exiting a session.

Incognito conversations end if the user closes the app or locks their phone. The chats are text-only, with users unable to upload or generate images. Users must confirm their age to use the mode because Meta does not allow users under 13.

The incognito mode includes safety features that steer users toward helpful information or refuse to answer harmful topics and may stop interacting. It will roll out to WhatsApp and the standalone Meta AI app over the next few months. The feature is based on WhatsApp's Private Processing infrastructure that debuted a year ago.

Incognito Chat uses Meta's Muse Spark model released last month. Meta AI was added to WhatsApp last year. In May 2025, Meta AI reached a billion users across Meta's apps, Mark Zuckerberg said. Will Cathcart, Meta’s head of WhatsApp, said the feature addresses user demand for privacy on sensitive subjects.

“We’re starting ask a lot of meaningful questions about our lives with AI systems, and it doesn’t always feel like you should have to share the information behind those questions with the companies that run those AI systems,” he said. Cathcart said people wanted private conversations with AI on sensitive subjects including health, relationships and finances.

Cathcart described the technology as 'the equivalent' of end-to-end encryption.

Alice Newton-Rex, VP of Product at WhatsApp, highlighted growing use of AI for private matters. “People are starting to use AI for everything, including some of their most private thoughts, whether that’s tackling financial or health questions, or for advice on how to respond to a tricky message from a friend or a colleague.

We think it’s really important to give people the ability to ask these questions as privately as possible,” she said.

Mark Zuckerberg said Meta AI is the first major AI product where there is no log of conversations stored on servers. He wrote that “Incognito Chat handles all AI inference in a Trusted Execution Environment that ensures your messages are not accessible to us. The conversations on your phone also disappear when you exit the session.

Zuckerberg wrote that he is proud that Meta Superintelligence Labs is the first lab to deliver private AI. Meta shares slid as its plan to spend billions more on AI spooked investors. Meta is on track to spend $145 billion on AI infrastructure in 2026, according to Susannah Streeter.

WhatsApp has blocked other AI chatbots from being accessible on its systems. Incognito Chat will be subject to expert oversight so third parties can verify the code. Prof Alan Woodward, a cyber security expert at Surrey University, said there was a low risk of compromising WhatsApp's existing security by introducing a second system.

Woodward raised concerns about lack of accountability if AI causes harm because chats disappear. He noted that disappearing messages which cannot be retrieved by the user or by Meta would mean it would be impossible to find evidence whether somebody's chats led to harm, death or suicide.

Key Facts

Incognito mode processes chats in Trusted Execution Environm
Messages are not accessible to Meta, are not saved by default, disappear on session exit or app close or phone lock, and are text-only with mandatory age confir
Meta plans $145 billion AI infrastructure spend in 2026
Spending announcement coincided with share slide as investors reacted to scale of investment
Feature includes third-party verification and safety guardra
Incognito Chat subject to expert oversight; built-in features refuse harmful topics; WhatsApp blocks other AI chatbots

Story Timeline

4 events
  1. 2026-05-13

    Meta announces rollout of incognito mode for WhatsApp and standalone Meta AI app

    4 sourcesMeta · ABC News · BBC News · TechCrunch
  2. Last month

    Meta releases Muse Spark model used to power new incognito chats

    1 sourceAlice Newton-Rex
  3. May 2025

    Meta AI reaches one billion users across Meta's apps

    2 sourcesMark Zuckerberg · BBC News
  4. One year ago

    Meta AI added to WhatsApp; WhatsApp's Private Processing infrastructure debuts

    3 sourcesBBC News · Wired · TechCrunch

Potential Impact

  1. 01

    Users gain ability to discuss health, finances and relationships with AI without logs stored on company servers

  2. 02

    Meta positions itself as first major lab delivering private AI while continuing heavy capital expenditure on infrastructure

  3. 03

    May accelerate adoption of AI chat features among WhatsApp's billions of users seeking privacy

  4. 04

    Cybersecurity experts note potential accountability gap if harmful AI advice occurs in disappearing chats

Transparency Panel

Sources cross-referenced4
Framing risk65/100 (moderate)
Confidence score98%
Synthesized bySubstrate AI
Word count544 words
PublishedMay 13, 2026, 2:00 PM
Bias signals removed2 across 2 outlets
Signal Breakdown
Loaded 1Speculative 1

Related Stories

VNET Sells 38% Stake to PJ Millennium Investors for $1.44 per ShareBenzinga
ai1 hr agoFraming65Framing risk65/100Rewrite inherits heavy consensus framing from sources by burying the VNET deal facts under lengthy speculative AI-crash commentary and mismatched defense-contract criticism, creating lede misdirection and valence skew.Click to jump to full framing analysis

VNET Sells 38% Stake to PJ Millennium Investors for $1.44 per Share

VNET Group Inc. announced a share purchase agreement with PJ Millennium I Limited and PJ Millennium II Limited for nearly 650 million shares. The transaction, valued at approximately $8.69 per ADS, is scheduled to close in Q4 2026 pending shareholder approval.

Cnbc
Benzinga
Forbes
Business Insider
Responsible Statecraft
+1
6 sources
SoftBank Reports $46 Billion Vision Fund Gain Driven by OpenAI StakeJapan Times
ai5 hrs agoUpdated

SoftBank Reports $46 Billion Vision Fund Gain Driven by OpenAI Stake

SoftBank posted a $46 billion gain at its Vision Fund for the financial year ended March, driven mainly by its stake in OpenAI. The Japanese company reported net income of ¥1.83 trillion in its fiscal fourth quarter, far exceeding analyst estimates. The performance has increased…

Cnbc
IN
Japan Times
Benzinga
5 sources
Walmart Reorganizes Technology Teams, Cutting or Relocating About 1,000 Corporate RolesNew York Post
ai1 hr agoDeveloping

Walmart Reorganizes Technology Teams, Cutting or Relocating About 1,000 Corporate Roles

The retail giant outlined the changes in a memo sent Tuesday by executives Daniel Danker and Suresh Kumar. The moves follow a review of overlapping teams after Danker was hired last year to lead global AI acceleration efforts. Walmart said the restructuring relates to organizatio…

New York Post
1 source