Meta warns users to 'avoid sharing personal or sensitive information' in its AI app

Engadget

6 hours ago

Meta seems to have finally taken a small step to address the epidemic of over-sharing happening in the public feed of its AI app. The company has added a short disclaimer that warns users to "avoid sharing personal or sensitive information" to the "post to feed" button in the Meta AI app.
The change was first spotted by Business Insider , which labeled the app "one of the most depressing places online" due to the sheer volume of intimate, embarrassing and sometimes personally-identifying information Meta AI users were — apparently unwittingly — publicly sharing to the app's built-in "discover" feed. Though Meta AI doesn't share users' chat histories by default, it seems that many of the app's users were choosing to "share" their interactions without realizing it would make the voice and text chats visible to the public.
Last week, I found posts where users asked for advice on "improving bowel movements" and inquiring whether a relative could be liable for their employer's unpaid taxes. Another user desperately added "keep this private" to his public posts in an apparent attempt to hide his embarrassing chats after the fact. These types of strange public interactions have been happening since the Meta AI app rolled out in April, but received renewed attention last week after social media users began posting about all of the weird conversations that were visible in the app's "discover" feed.
Privacy experts criticized Meta, noting that most other mainstream AI chatbots don't include a social, publicly-visible feed. "If a user's expectations about how a tool functions don't match reality, you've got yourself a huge user experience and security problem," Rachel Tobac, a security expert who has previously partnered with Meta, observed last week. "Humans have built a schema around AI chat bots and do not expect their AI chat bot prompts to show up in a social media style Discover feed — it's not how other tools function." The Mozilla Foundation also urged Meta to change the app's design. "Meta AI's app doesn't make it obvious that what you share goes fully public," it wrote in a statement last week There's no clear iconography, no familiar cues about sharing like in other Meta apps."
Now, the company has apparently taken note. With the change, choosing to share a Meta AI interaction publicly prompts the warning seen above, though it only seems to appear on the first share. "Prompts you post are public and visible to everyone," it states. "Your prompts may be suggested by Meta on other Meta apps. Avoid sharing personal or sensitive information."
As Business Insider notes, the app's public feed also seems to no longer feature text exchanges other users have shared with the app, only AI-generated images and video. It's unclear if that's a permanent change, or the result of the recent negative attention the app's received. We've reached out to Meta for more information and will update if we hear back.
In the meantime, if you've found yourself the victim of unintended public posts in the app, you can remove them by tapping on your profile in the top right corner of the app, heading to Data & Privacy -> Manage your information -> Make all public prompts visible only to you and selecting "apply to all."