The current circulating in WhatsApp groups this week claims that Meta AI, the application’s artificial intelligence, can read group messages and private conversations if administrators do not immediately activate “advanced privacy”. The message calls for urgent action and scares those who receive it.
The allegation is false. WhatsApp’s end-to-end encryption continues to work normally and prevents Meta or any AI from accessing the content of conversations without the user explicitly interacting with the tool.
What exactly does the chain say?
The text shared in a chain warns that the AI can “open group messages, see phone numbers or even access personal cell phone data.” It directs group administrators to activate advanced privacy settings to supposedly block this access.
Fact-checking experts, such as the Austrian website Mimikama, deny the content. The current mixes two different features of the app and creates a risk that does not exist.
How Meta AI really works
Meta AI appears as a blue-purple circle in the corner of the screen and has been available since 2025. It only processes information when the user triggers it directly — for example, by sending it a message or using the @Meta AI command in a group.
Outside of these interactions, conversations between people remain protected by end-to-end encryption. Meta’s “Private Processing” technology ensures that, even when used, content is treated in a temporary and encrypted form, without permanent storage by the company.
What Advanced Privacy Really Does
The “advanced chat privacy” option exists and serves to protect sensitive chats. When activating it:
- Prevents participants from exporting the conversation;
- Blocks the automatic download of media;
- Restricts the use of certain AI functions by other users in the chat.
It has no relation to Meta AI’s alleged access to messages. Its objective is to give more control against the actions of other group participants, not against the company.
Why does this rumor keep coming back?
Similar messages circulated in previous waves, with slight variations in date or detail. The appeal comes from the use of real terms like “Meta AI” and “advanced privacy”, which gives false credibility. Fact-checkers from outlets like G1, Snopes and Mimikama have already denied similar versions several times.
Practical tip: It is not necessary to activate any extra configuration because of this current. Anyone who does not interact with Meta AI has no reason to worry. For those who use it, simply avoid sharing sensitive data directly with AI.