Threads users can now chat with Meta AI in their DMs

Meta wants you to talk to its AI in private. On Monday the company announced that Meta AI is now available inside Threads direct messages, letting users share posts, images, links, and videos with the chatbot and ask follow-up questions without ever leaving the app. The move is straightforward: make the assistant unavoidable, make the switch to a competitor feel like friction, and call it a feature.

The strategy is transparent. Threads already let select markets interact with Meta AI in public replies, mimicking Grok on X. But public feeds are messy. People mock bad answers. They screenshot hallucinations. They mute the bot. By moving the conversation into DMs, Meta removes the audience. No performative failures. No public backlash. Just a user and a model in a quiet room, where the stakes feel lower and the habit forms faster.

Meta AI already sits in DMs across Facebook, Instagram, and WhatsApp. Threads was the holdout. Now the circle closes. The company frames this as giving users "an easier way to talk to an AI chatbot." The real goal is simpler: reduce the reasons to open ChatGPT or Gemini. Every query answered inside the Meta stack is a query that doesn't feed a rival's training data, doesn't build a rival's user graph, doesn't threaten the moat.

The global rollout starting Monday contrasts with the cautious public feed test. Meta says it's "considering feedback before expanding availability more broadly" for in-feed AI replies. Translation: the product isn't ready for prime time. The company knows it. That's why it offers mute controls — mute @meta.ai, hit "Not interested," hide replies on your own posts. These aren't power-user tools. They're escape hatches. Meta built them because it expects people to use them.

Thread's positioning as an X rival has always been ambiguous. It launched as a text-first companion to Instagram, then borrowed features from the platform it supposedly competes with. Now it borrows the AI layer too. The pitch: get information and recommendations without leaving the app. But the information comes from a model that famously struggles with recency, cites sources it can't verify, and defaults to bland consensus. The recommendations carry the weight of a company that monetizes attention, not accuracy.

Users who want fewer Meta AI replies can mute the bot. That sentence should worry Meta more than any competitor. A product people actively silence is a product that hasn't earned its place. The company calls it "feedback." It's rejection.

The deeper problem: Meta AI isn't differentiated. It runs on Llama 3, a capable open-weight model. But so do dozens of wrappers. The only moat is distribution — the billions of accounts across Facebook, Instagram, WhatsApp, and now Threads. Distribution is a real advantage. It's also a fragile one. If the experience disappoints, the switch cost is near zero. ChatGPT lives in a browser tab. Gemini lives in a pixel. The user who learns that Meta AI invents citations or flattens nuance will stop invoking it. The mute button makes that choice effortless.

Meta's bet is that convenience beats quality. That habit beats intent. That the path of least resistance becomes the path of only resistance. It's a bet the company has won before — with Stories, with Reels, with the feed algorithm itself. But AI is different. The output is legible. The errors are visible. The comparison is instant. A bad Reel scrolls past. A bad answer sits in the chat history, proof that the assistant isn't assisting.

The rollout will expand. The mute data will accumulate. Meta will learn how many users try the bot once, twice, three times, then silence it. That number will matter more than any engagement metric the company publishes. For now, the invitation stands: talk to our AI in your DMs. It's private. It's convenient. It's inside the walls. Whether it's useful — that's the variable Meta doesn't control.