In March 2025, WhatsApp introduced a deeply integrated feature: Meta AI, a chatbot powered by artificial intelligence. Visibly embedded on the app’s main screen as a blue-circled icon, this function cannot be removed or deleted. The launch has caused unease among privacy-conscious users, particularly in Germany, where digital rights are strongly protected by law. On social media, viral messages claim that Meta AI accesses all private chats and phone numbers. While Meta has denied these claims, trust in major tech companies remains fragile. This report is based on an analysis by G.Business, referencing Renewz.
What Meta AI Does — And When It Becomes Active
Meta AI does not activate automatically. It becomes operational only when a user enters the command @Meta in a chat window. Without this explicit interaction, the AI remains dormant. All WhatsApp chats are encrypted end-to-end by default, meaning only participants in a conversation can access its content. Furthermore, WhatsApp does not upload contacts without consent — phone numbers are only shared if users manually sync their address books with the app.
Technical Safeguards and How Data Is Handled
When Meta AI is used — for example, to summarize missed messages (a feature currently available only in the United States) — the system establishes a temporary, anonymized session with a cloud server. This is done using a method known as Oblivious HTTP (OHTTP), which routes requests through intermediary servers to mask the origin and content of the data. The cryptographic key used in the session is not accessible to Meta, according to the company. All processing takes place in Trusted Execution Environments (TEEs) — secure, isolated processing units that prevent unauthorized access even from Meta itself.
Legal Protections for WhatsApp Users in Germany
Under German law, WhatsApp is classified as a telecommunications service provider. This means that it is subject to the country’s telecommunications secrecy law, regulated under § 3 of the Telecommunications-Digital Services Data Protection Act (TDDDG) and Article 10 of the Basic Law (Grundgesetz).
Key legal protections include:
- The confidentiality of communications is guaranteed by constitutional law
- Unauthorized data access is a criminal offense under § 206 of the German Criminal Code (StGB)
- Content and metadata of user chats are protected by federal law
- Meta, like any other provider, is legally restricted from reading or analyzing chat content without explicit permission
How to Block Meta AI in Individual Chats
Although Meta AI cannot be removed from WhatsApp entirely, the company offers a feature called "Advanced Chat Privacy", launched in April 2025. This allows users to block Meta AI on a per-chat basis.
Step-by-step instructions to block Meta AI:
- Open the individual or group chat in which you want to disable Meta AI
- Tap on the name of the contact or group (not the three-dot menu)
- Select "Advanced Privacy"
- Activate the toggle for "Block Meta AI"
- After this, if someone tries to use
@Metain that chat, WhatsApp will display: "This feature is not available in this conversation"
Please note: this must be done manually for each chat and is not retroactive.
What Does the Privacy Feature Disable
The "Advanced Chat Privacy" setting disables more than just Meta AI. It also prevents the chat content from being exported or the media from being automatically saved to the device gallery.
| Feature | Effect when enabled |
|---|---|
| Meta AI in chats | Disabled — cannot be activated via @Meta |
| Chat export | Blocked — chat cannot be shared or saved |
| Media access | Blocked — photos/videos not saved locally |
| Cloud processing | Not available in blocked chats |
Who Should Consider Using the Block Feature
The feature is not necessary for everyone, as Meta AI does not operate automatically. However, it is particularly useful in contexts where data privacy is critical.
Use cases include:
- Professionals handling confidential data (e.g., lawyers, doctors, journalists)
- Parents managing children's group chats
- Administrators of large WhatsApp groups with sensitive communications
- Anyone who wants to guarantee that AI is never activated in a conversation
Meta AI is now a permanent visual feature in WhatsApp, but it does not operate in the background or read messages without user input. It cannot be uninstalled, but it can be blocked per chat. Between end-to-end encryption, legal protections under German constitutional and criminal law, and technical safeguards such as OHTTP and TEE environments, WhatsApp users retain significant control over their private data. The “Advanced Chat Privacy” function provides an additional layer of reassurance for those who want to ensure their conversations remain AI-free — and fully under their control.
Stay connected for news that works — timely, factual, and free from opinion — and insights that matter now: Grok 4.20 by Elon Musk: What’s Known About the New AI Release – Advantages, Risks.