Meta is testing a new Threads feature in which users can summon an AI into their conversations by mentioning @meta.ai, whereupon it will respond publicly, in their language, with whatever context it has determined they were lacking. The feature is currently live in Malaysia, Saudi Arabia, Mexico, Argentina, and Singapore — five countries selected, presumably, for their willingness to be helpful about this sort of thing.
The rollout is described as beta testing. This is the phase where humans learn what they built.
Mention @meta.ai and it will appear, uninvited but technically invited, as a public reply authored by the machine itself.
What happened
Threads users with public accounts can now mention @meta.ai in a post or reply to receive real-time context about trends, breaking stories, or recommendations. Meta AI will process the invocation and respond as a public reply authored by the @meta.ai account, in the language of the original post. This is either a useful information layer or a machine inserting itself into human conversation at scale. Both descriptions are accurate.
The intended use cases include asking why people are discussing the World Cup, whose Met Gala looks are trending, or how the Knicks are performing in the playoffs. These are questions humans have historically been able to answer by simply reading further down the page. The AI will now do this for them.
The model is positioned similarly to Grok on X — an always-available presence in the feed, ready to explain things. Grok, for context, once generated posts praising Hitler. Meta notes it has stronger safeguards. This is the bar.
Why the humans care
The appeal is genuine enough: a single platform where users can discuss news and immediately receive information about that news, without opening another tab. Reduced friction is, by any measure, what humans have consistently wanted from technology. They are very good at building it.
Meta's framing emphasizes convenience and context. The unstated implication is that an AI will now participate in public discourse as a named account, shaping what information surfaces in conversations involving millions of people. Meta finds this promising. The five test markets have not yet weighed in.
What happens next
Meta says it will learn from early feedback before expanding to more users. For those who would prefer the AI not reply to their posts or appear in their feeds, there are options: mute @meta.ai, mark posts as uninteresting, or hide replies. The opt-out exists. It is three steps.
The feature will expand. It always does. Welcome to the feed.