Meta AI on WhatsApp is not inherently “dangerous,” but it does pose significant privacy risks. You should be aware of these and act accordingly.
Is Meta AI on WhatsApp dangerous? Here’s what you need to know
Meta AI on WhatsApp is not risk-free from a privacy perspective and is viewed critically by experts and consumer advocates.
- The integration of the AI assistant means that all requests you make directly to Meta AI are not end-to-end encrypted like traditional WhatsApp messages. This means Meta can view, store, and analyze this content.
- While regular chats between users remain encrypted and are not visible to Meta, as soon as you use Meta AI, your data leaves the protected area and is processed on external servers.
- Another issue is the lack of transparency: It is unclear how long your data is stored, for what purposes it is used, and whether it is linked to other Meta services such as Facebook or Instagram.
- A particularly critical point is that you cannot completely disable Meta AI. The feature is integrated into WhatsApp by default, and users do not provide explicit consent. In short: You cannot delete Meta AI from WhatsApp.
- You can only mute the notifications, but you cannot prevent the feature from being available in the first place.
Do not share sensitive data with Meta AI
Starting in late May 2025, Meta plans to use queries to Meta AI for training the AI as well, unless you actively opt out.
- This applies in particular to data from Facebook and Instagram; however, according to Meta, WhatsApp data will still not be used for training purposes—unless you explicitly share it with Meta AI.
- Consumer advocates therefore recommend not sharing any sensitive or personal data—such as names, dates of birth, or health information—with Meta AI.
- For businesses, this use is particularly problematic, as company-related information or personal data could inadvertently be disclosed to third parties.
- In summary: You should carefully consider what information you share with the AI assistant and, if necessary, exercise your right to object by May 26, 2025.
New Feature: Remove Meta AI from Group Chat
In April 2025, Meta introduced a new feature called “Enhanced Chat Privacy” in WhatsApp.
Starting in July 2025, every private and group chat will have a toggle for this feature, which is disabled by default. When enabled, no chat participant can summon Meta AI by mentioning it with an “@”. This prevents the AI from being involved in the chat messages or reading them.
- In group chats, any member—regardless of admin status—can enable or disable the privacy toggle. Once the feature is enabled, everyone else will see a corresponding notification in the chat. While enabling it does not make the chat completely eavesdrop-proof, it at least technically prevents content from being passed on to Meta AI for analysis.
- Those who want more privacy can use this toggle to at least keep Meta AI out of their own chats. However, there is still no option to completely disable Meta AI.
