What Meta Is Actually Testing Right Now

If you've spent any time on X watching someone tag Grok under a viral post going "wait, is this real???" — buckle up, because Threads is about to give you that same experience.

Meta is testing a new feature that gives its AI chatbot a dedicated account on Threads: @meta.ai. Users can tag it directly inside posts and replies, and the bot jumps in publicly with context, recommendations, or information about whatever's being discussed. It's essentially AskGrok, but dressed in Meta's colors.

Right now, the feature is in early beta. Only users in Malaysia, Saudi Arabia, Mexico, Argentina, and Singapore can access it. But the way early betas tend to go, it's only a matter of time before it rolls out wider.

How the @meta.ai Bot Actually Works on Threads

The mechanic is simple. Tag @meta.ai in any post or reply, ask it something — "why is everyone talking about the Met Gala right now?" or "how are the Knicks doing in the playoffs?" — and it responds. Publicly. Right there in your thread, for everyone to see.

And here's a nice detail: it replies in whatever language the original post was written in. For a platform with a genuinely global audience, that's a thoughtful touch.

The bot runs on Meta's latest Muse Spark model — the same one being rolled out across WhatsApp, Instagram, Facebook, and Messenger. So this isn't a one-off experiment. It's part of a much bigger push.

The Private Version on WhatsApp

On WhatsApp, Meta is testing something similar but different in one key way: it's private. The feature is called "side chats," and you can ask the bot about your group conversation with only you seeing the response. No public replies, no chaos. Just quiet AI assistance tucked into the background.

That contrast matters. Threads is built for public discourse. WhatsApp is built for private conversation. The same AI behaves very differently depending on which room it's in.

Why People Are Already Worried — And Why That's Not Surprising

Here's the thing: Grok on X has given us a pretty clear preview of what can go wrong when you put a public AI bot in the middle of millions of conversations. Grok drew serious backlash after generating non-consensual sexualized images of real people. The fallout eventually pushed X to add a toggle letting users block Grok from editing their photos — though critics noted even that fix fell short.

Meta has historically kept tighter guardrails on its AI than X has with Grok. That's worth acknowledging. But tighter guardrails don't mean no problems. Putting any AI chatbot this visible on a public social platform opens the door to the kind of user-driven chaos that X has spent months trying to explain away.

And Threads users wasted no time proving that point.

The Blocking Problem Nobody Saw Coming (Or Did They?)

Within hours of @meta.ai going live, users discovered something that sent the platform into a spiral: you can't block it. The bot is there. It can be tagged into your thread by anyone. And you have no way to opt out.

That single design decision triggered over a million angry posts on the platform. Not a slow burn — an immediate, overwhelming backlash. People don't love the idea of an AI bot being able to show up in their conversations without their permission, and honestly? That reaction makes complete sense.

The Bigger Picture Here

This isn't just about one feature on one app. It's about a pattern. AskGrok normalized the idea of a public AI participating in social media conversations. Now Meta is doing the same thing on Threads. The playbook is identical: tag the bot, get a public response, watch what happens next.

The difference — maybe — is that Meta has a track record of being more careful. But "more careful than X" is a pretty low bar. And the blocking issue suggests that, at least at launch, some basic user controls weren't fully thought through.

There's a real tension here between making AI feel seamlessly woven into your social experience and making sure people still feel like they have some control over their own feeds and conversations. Right now, on Threads, it seems like the seamless integration part got prioritized. The control part... less so.