Image source: Meta AI WhatsApp on X: https://x.com/WhatsApp/status/1781010491961229360
Understanding Meta AI on WhatsApp: the basics when it comes to keeping your messages private
Meta has integrated its AI, called Meta AI, into WhatsApp. This lets you ask questions, generate images, and get information directly within your chats. However, if you are using WhatsApp for business with clients or family and discussing private or confidential matters, is using Meta AI in those chats going to give you the same level of privacy as normal WhatsApp messages?
What happens when you use MetaAI in WhatsApp messages?
When you use features like asking Meta AI a question or generating an image, you’re directly interacting with Meta’s servers. This means information is being sent to Meta.
What Meta collects (and what they can’t)
Meta’s official statement clarifies what they receive when you use Meta AI:
- Prompts and AI messages: When you type a question or request (a “prompt”) to @Meta AI, or receive a response, that text is sent to Meta. This is necessary for the AI to work.
- Feedback: If you provide feedback on the AI’s responses (e.g., “that was helpful” or “that wasn’t right”), that feedback is also sent to Meta.
- Performance information: Meta also collects data on how well the AI is performing, which helps them improve it.
- Meta AI can remember details you choose to share to give you more relevant responses. Whenever Meta AI saves a detail, you’ll see Memory updated below the prompt you sent Meta AI. You can tap Memory updated or Meta AI > Memory to view or delete details. You can delete details that Meta AI saves about you at any time.
Meta scan ONLY read messages that directly mention @Meta AI, or that users choose to share with Meta AI. They can not read any other messages in your personal chats.
The crucial role of end-to-end encryption
Here’s the most important part for your privacy:
End-to-End Encryption: WhatsApp’s core messaging feature uses end-to-end encryption. This means that your regular messages and calls are scrambled so that only you and the people you’re chatting with can read or hear them. Not even WhatsApp or Meta can access them.
Meta AI Interaction is Different: When you use Meta AI, you’re stepping outside of the standard encrypted messaging. You’re actively engaging with Meta’s servers, so those specific interactions are not protected by end-to-end encryption.
How to protect your privacy when using Meta AI
While you can’t have the same level of privacy as standard encrypted messages when using Meta AI, here’s how to minimise risks:
Be mindful of what you share:
- Avoid sharing highly sensitive or confidential information in prompts to Meta AI. Those prompts are sent to Meta.
- Don’t include personal data like your social security number, bank details or extremely private conversations.
Use Meta AI only when necessary
If you’re discussing something very sensitive, stick to regular WhatsApp messages within your group, which are end-to-end encrypted. Only use the Meta AI when you need to or outside of WhatsApp.
Understand the Terms and Privacy Policy
Meta encourages you to read the Meta AIs Terms of Service and the Meta Privacy Policy. While they can be lengthy, they explain how your data is used.
Limit group usage
While the AI will only respond to direct mentions, if a group chat is discussing very sensitive topics, it is best to refrain from using the AI within that group.
Meta AI is a tool, not your confidant
Meta AI is a tool provided by a company. It is not a person and it does not have the same level of confidentiality as a conversation between trusted individuals due to the nature of how it functions.
Think of it like this: Your regular WhatsApp messages are like letters sealed in an envelope. No one but the recipient can open them.
Using Meta AI is like writing a postcard. Those who have access to the “post office” (Meta servers) might read what’s on the postcard.
Confidence level
According to Meta, you can be 100% confident that your regular WhatsApp messages are private due to end-to-end encryption. However, when using Meta AI, you must accept that Meta will have access to the information you share with it.
Where is MetaAI available?
Meta’s AI chatbot Meta AI is available in the United States and 21 other countries: Argentina, Australia, Cameroon, Canada, Chile, Colombia, Ecuador, Ghana, India, Jamaica, Malawi, Mexico, New Zealand, Nigeria, Pakistan, Peru, Singapore, South Africa, Uganda, Zambia and Zimbabwe.
According to CNet, “Though Meta AI isn’t available in the EU, the company says it might later join the EU’s AI Pact. The AI Act requires companies to provide “detailed summaries” of the data used to train their models — a requirement Meta has been hesitant to meet, likely due to its history with data privacy lawsuits.”
Whatever you do, do not purchase Meta glasses while on holiday in countries where they have yet to adopt MetaAI. Otherwise, you will not be able to use them when you get back home, for example in the US or UK. This could be a common error for people who are online shopping while on holiday in the European Union, for example.
According to a Reddit discussion and Meta response, the glasses will forever be linked to the country of purchase, so resetting your apps and VPN won’t work. You would just have to wait until the country adopted MetaAI before you could use them.
The AI assistant is available on Facebook, Instagram, WhatsApp, Messenger apps, and on the Meta.ai website.
Did you find this article useful?
Then join thousands of other informed freelancers by signing up for the FREE Freelance Informer newsletter to receive news, expert views and business, lifestyle, money and career tips every week:
