Meta Ai adds an additional step to avoid embarrassing areas

This delicate conversation on your wife, or a call for help on your daughter – or the strange pocket dial which recorded a conversation with your friend – are all less likely to be accidentally finding the public flow of Meta’s Ai App.

Mark Zuckerberg The company now offers users a new pop-up warning before sharing something at Meta AI’s “Discover” flow.

This comes after Business Insider reported last week that the flow was full of personal information that seemed to have been accidentally shared – things like an article that showed a parent asking the AI ​​of advice on a capricious girl, the audio of a pocket dial where someone chatted to colleagues of his working hours and the text of an employee in search of a legal opinion on their employer.

Meta I did not answer my request for comments on Monday when I asked questions about the new pop-up.

After the story of BI last week, the BBC, The Washington Post,, Cable,, TechcrunchAnd other media have organized their own stories about the apparently accidental extension.

The new warning message, which was visible on Monday, appears after hitting “sharing” on a conversation in the Meta AI application. The message indicates: “The prompts you publish are public and visible for everyone. Your prompts can be suggested by META on other meta apps. Avoid sharing personal or sensitive information.”

After this warning, the “Poster to feed” button is disabled until you press again in the middle of the screen. Only SO Can you publish the public discover Feed.

Meta Ai’s flow had strangely personal information

Meta Ai, the AI ​​autonomous chatbot, has its “discover” flow so that people share images they have created with AI or to publish their conversations with the bot. These cats are not public by default; You have always had to click on “share” then “publish” to send your cats to the discovery flow.

Obviously, it seems that some users did not understand what they were doing in the previous system. (A man I spoke last week said that he had not wanted to publish his conversation with the chatbot on the repair of the car.)

When the application was launched at the end of April, I wrote on the problem of accidental publications on Meta Ai. By checking last week, I noticed that the problem seemed even worse – there were now many more messages that seemed to be clearly shared by mistake. Some of them included intimate personal problems, such as asking for helping a letter from a judge on child care, medical problems and dilemmas of relationships.

People on x Also have found examples of people publishing tax information, or seeming not to understand what they used, like a person who asked him what would happen if they would put a deep heat cream on their intimate pieces, but then asked the bot to keep the private conversation (on a public flow). Hmm.

Meta’s new pop-up warned people before publishing

Now, of course, it is quite possible that some of these embarrassing pure posts are secretly intentional. (The person who asked Meta I have to give him a variety of images of celebrities doing strange things, including Ana de Armas tidying Spider-Man in bed, may have wanted people to see his business.)

Another change in the discovery flow of Meta Ai that I noticed on Monday: these are now almost exclusively images. Last week, and the times that I had checked on the application since it started at the end of April, the flow had a heavy pepper of messages based on text and audio, like these apparently accidental pocket dials.

Meta also did not respond to BI if the apparent adjustment of the discovery flow was intentional – as if things other than images have been buried or disappeared – or if change is a potential natural consequence of the new additional tap necessary to share something to the public flow. Or if it’s something else.

I will update this story if Meta is coming back to me, but for the moment, I’m happy not to have to close my eyes, to cringe, to start (as much) when I scroll the Meta AI discovery flow.