When Meta rolled out its standalone AI chatbot app in April, it didn’t make much noise. The tool, aimed at light conversation rather than heavy lifting, slipped into app stores with little fanfare. But now, weeks later, what was meant to be a glimpse into AI’s friendly face has taken a turn, and not necessarily for the better.
One of the app’s standout features is the so-called "Discover" feed. On paper, it’s a showcase, a stream of chats, prompts and images from users who chose to share them. In practice, it feels more like a digital confessional. Scrolling through it, you don’t just find holiday recipe ideas or cartoon requests. You find people pouring their hearts out to a machine, and, unintentionally, to the world.
Among the sea of casual requests, there are moments that stop you in your tracks. As per BI findings, a man asks for a prayer to start his morning. Another wants help writing to a judge about child custody. Someone else is trying to write a letter to their local club after being suspended. It’s not just the topics that feel out of place, but the sense that some users may not realize they’ve made their conversations public.
The app doesn’t make posts public by default. Meta says it takes a deliberate tap of “Share” and then “Post” before anything goes live. Even so, some of what’s appeared, including personal phone numbers and email addresses, suggests that at least a few users might have clicked without fully understanding the consequences.
There’s more. If someone uses voice chat with Meta’s bot and then shares that exchange, it shows up in the feed too, audio and all. In a few recordings, background chatter gives the impression that the user had no idea they were broadcasting. In one, two people discuss a work schedule while Meta’s assistant tries to join in. Another sounds like someone’s phone had been nudged in a pocket, with the bot occasionally piping up into thin air.
The unsettling part isn’t just that people are sharing intimate moments with AI. It’s that their real Facebook or Instagram profiles are attached to these posts. A stranger scrolling through the feed can tap on a name and, in some cases, pull up someone’s full social media footprint. At least one user, contacted after posting a car repair query, said they hadn’t meant to share their chat publicly at all.
To be fair, not every post is so loaded. Plenty of users are simply asking for help with dinner plans, looking for writing prompts, or generating playful images. Still, the overall experience feels patchy — a mixed bag of the mundane, the bizarre, and the worryingly personal. It’s like overhearing bits of private conversations in a crowded room where no one realizes their mic is on.
Meta has made no secret of its ambitions in AI. The company is pouring resources into the space, eyeing advances that could eventually outpace human intelligence. But in the short term, the public-facing side of that effort — the part ordinary users can actually see — feels half-baked.
The Discover feed was meant to highlight creativity and connection. Instead, it gives the impression that many users don’t quite know what they’ve signed up for. If Meta wants the public to see its AI as the future, it might need to do a better job showing how that future protects, rather than exposes, the people using it.
Image: DIW-Aigen
Read next:
• Edit Your Video by Editing the Words: Microsoft Adds Transcript Tool to Clipchamp
• Invisible Emissions From Digital Mics Could Let Strangers Hear You Through Concrete
One of the app’s standout features is the so-called "Discover" feed. On paper, it’s a showcase, a stream of chats, prompts and images from users who chose to share them. In practice, it feels more like a digital confessional. Scrolling through it, you don’t just find holiday recipe ideas or cartoon requests. You find people pouring their hearts out to a machine, and, unintentionally, to the world.
Among the sea of casual requests, there are moments that stop you in your tracks. As per BI findings, a man asks for a prayer to start his morning. Another wants help writing to a judge about child custody. Someone else is trying to write a letter to their local club after being suspended. It’s not just the topics that feel out of place, but the sense that some users may not realize they’ve made their conversations public.
The app doesn’t make posts public by default. Meta says it takes a deliberate tap of “Share” and then “Post” before anything goes live. Even so, some of what’s appeared, including personal phone numbers and email addresses, suggests that at least a few users might have clicked without fully understanding the consequences.
There’s more. If someone uses voice chat with Meta’s bot and then shares that exchange, it shows up in the feed too, audio and all. In a few recordings, background chatter gives the impression that the user had no idea they were broadcasting. In one, two people discuss a work schedule while Meta’s assistant tries to join in. Another sounds like someone’s phone had been nudged in a pocket, with the bot occasionally piping up into thin air.
The unsettling part isn’t just that people are sharing intimate moments with AI. It’s that their real Facebook or Instagram profiles are attached to these posts. A stranger scrolling through the feed can tap on a name and, in some cases, pull up someone’s full social media footprint. At least one user, contacted after posting a car repair query, said they hadn’t meant to share their chat publicly at all.
To be fair, not every post is so loaded. Plenty of users are simply asking for help with dinner plans, looking for writing prompts, or generating playful images. Still, the overall experience feels patchy — a mixed bag of the mundane, the bizarre, and the worryingly personal. It’s like overhearing bits of private conversations in a crowded room where no one realizes their mic is on.
Meta has made no secret of its ambitions in AI. The company is pouring resources into the space, eyeing advances that could eventually outpace human intelligence. But in the short term, the public-facing side of that effort — the part ordinary users can actually see — feels half-baked.
The Discover feed was meant to highlight creativity and connection. Instead, it gives the impression that many users don’t quite know what they’ve signed up for. If Meta wants the public to see its AI as the future, it might need to do a better job showing how that future protects, rather than exposes, the people using it.
Read next:
• Edit Your Video by Editing the Words: Microsoft Adds Transcript Tool to Clipchamp
• Invisible Emissions From Digital Mics Could Let Strangers Hear You Through Concrete
