‘WhatsApp is allowing AI to read chats,’ claims Paytm founder, advising users to enable this setting

Paytm founder Vijay Shekhar Sharma took to X and remarked that WhatsApp’s advertised “end-to-end encryption” feature is not as secure as it claims to be.

Sharma said that the messaging and calling app is now allowing artificial intelligence (AI) to “read chats” if one is added to a WhatsApp group, urging users to “enable” a setting to “block” WhatsApp from doing so. 

“If you are part of a WhatsApp group. Today onwards, WhatsApp is allowing AI to read chats,” the Paytm boss claimed on X.

How to protect AI from reading your chats

Sharma encouraged enabling a setting to block WhatsApp from doing so. If you are added to any group on WhatsApp and go to “group info”, you will see an option for “advanced chat privacy”.

It is marked off by default. If you turned it on, people in the group won’t be able to do three things: 

1- Save media to their device gallery automatically. 

2- Use AI to summarise chats by mentioning “@MetaAI”. 

3- Export the chat. 

The advanced chat privacy of WhatsApp says: “Limit how messages and media from this chat can be shared outside of WhatsApp.”

It adds, “Your personal messages are protected with end-to-end encryption even if you don’t turn on advanced chat privacy. No one outside of the chat, not even WhatsApp or Meta, can read, listen to, or share them.”

Reacting to Sharma’s post, WABetaInfo said that it is “inaccurate”. 

It explained, “Only messages that mention Meta AI are shared with it, and you can check if Meta AI has read your message by opening the Message Info screen. If you choose to enable Advanced Chat Privacy, you won’t be able to use some AI features, but this doesn’t mean your messages are automatically shared with Meta AI.”

🚨IMP If you are part of WhatsApp group. Today onwards WhatsApp is allowing ai to read chats. ‼️‼️
So enable this setting to block it. pic.twitter.com/VhJomO4Msd

— Vijay Shekhar Sharma (@vijayshekhar) August 19, 2025

Netizens react

However, social media users lauded Sharma for the useful post.

One X said, “Privacy is completely Moh Maya now! Everything is open via one or another tool. Normal people hardly care/know. That’s exactly why data is the new fuel!”

“Thanks for the heads up,” said another user.

A third user commented, “One can never trust Meta.”

Another user wrote, “Crazy that one will have to do this for every group and not in one go. Definitely the intent is to train LLMs.”

Note from WhatsApp

A WhatsApp spokesperson reached out to Startup Pedia today and confirmed that the instant messaging app is always protected by end-to-end encryption.

The WhatsApp spokesperson said: "Your personal messages on WhatsApp are always protected by end-to-end encryption so that only you and the people you chat with can read or share them. Meta AI on WhatsApp is completely optional and, if you choose to message or invoke Meta AI in your chats, it can only read what you share with it."

Also read: NHAI sells over 5 Lakh FASTag annual passes, generates ₹150 Crore in 4 days (startuppedia.in)