Meta AI App is reportedly exposing your private chats

Meta's new AI app is facing backlash after users discovered their private chats may be publicly accessible. Experts urge Meta to fix its privacy design.

Of late, we’ve all started talking to AI chatbots as if they were our friend. We share our problems, ask for solutions and often upload personal data, hoping for advice, edits or insights. And let’s be honest: we all believed we were in a safe space. I know I did.

Meta AI App sparks privacy concerns over reports of leaked private chats.(Unsplash)

One such AI that many people have welcomed into their digital lives is the Meta AI app, Meta’s standalone chatbot that works across Facebook, WhatsApp and Instagram. Now, what if we told you this app has been reportedly exposing your conversations with the bot to the public?

You’d feel unsafe, uncomfortable, maybe even betrayed. Right? Well, that’s exactly what’s been happening.

Launched quietly in April, the Meta AI app includes a “Discover” feed, a public wall where user interactions with the bot can be shared. While posting is technically optional, the problem is that many users have no idea that their chats, including sensitive questions and audio, are being published for anyone to see, as per recent media reports. 

Users have no clue they are sharing their chats publicly
The Meta AI app features a prominent “Share” button, but there’s no clear warning, no privacy prompt and no obvious notification telling users that their messages could end up on a public feed. If you’re logged in with a public Instagram account, no matter how personal your AI searches are, they might be out there for anyone to see, as per a report by TechCrunch.

This has already led to a wave of oversharing of deeply sensitive posts from confessions about tax evasion and legal advice involving named individuals, to medical issues, court records and even full home addresses.

And it doesn’t stop with the app. Meta AI is now built into platforms like WhatsApp and Instagram. But unlike regular messages, AI chats aren’t protected by end-to-end encryption. 

A design flaw that feels like a breach of trust

Despite the storm, Meta hasn’t offered clear fixes or major design changes. It simply says nothing is shared unless you choose to share it, but the interface makes it all too easy to do just that without realising the consequences.

So far, the Meta AI app has clocked 6.5 million downloads, not huge by Meta standards, but big enough to fuel what feels like a growing PR crisis. One second, users are asking travel tips; the next, trolls are posting job applications alongside bizarre, meme-fuelled prompts.

Can users prevent this?

To prevent this, users must manually dive into settings, tap the profile icon and go to “App settings” > “Data & Privacy” > “Manage your information”, and toggle “Make all your prompts visible to only you.”

But this process is buried and far from intuitive. Whether Meta underestimated how people would use this feature or ignored the risks, the damage is underway.

Till the time the assumption of privacy remains an assumption, cybersecurity experts strongly advise against using the app for anything remotely sensitive