Meta has rolled out a Meta AI integration in Threads that replies publicly inside users’ feeds when mentioned, in a beta now live across five markets. Threads users who tag @meta.ai in a post or a reply get a public response from the AI account in the same language, covering trends, context or quick recommendations. Meta’s stated goal is to keep people inside Threads when they want a real-time explainer.
The Verge has since confirmed a separate detail that Meta did not lead with: the @meta.ai account cannot be blocked. Users can mute it, hide its replies or tap Not interested, but the standard Threads block button is not available. Australia is not in the beta, but the design call still lands here.
What Meta is actually testing
The integration is closest to how Grok works on X, according to TechCrunch. Threads users with public accounts can tag @meta.ai in any post or reply and the assistant will respond publicly, in the language the original post used. The beta launched in Malaysia, Saudi Arabia, Mexico, Argentina and Singapore. Meta has not committed to a broader rollout date.
Meta described the use case in plain terms: explaining a trend, surfacing recommendations, or filling in context on a breaking story without making someone leave the app. The replies appear as normal Threads posts attributed to @meta.ai, which means they are visible to anyone who can see the thread.
The block button that isn’t there
The Verge spotted a detail Meta didn’t flag at launch: there is no block option on @meta.ai. Users get three controls instead. They can mute the account, mark specific replies as Not interested, or hide Meta AI replies altogether from their feed.
On every other Threads account, blocking is standard. Removing the option here makes @meta.ai a non-optional presence on the platform, even for people who never tag it. Meta points to a Not interested signal and content moderation guardrails, but the lopsided rules are what’s drawing most of the heat in the first 48 hours.

Why Australian businesses should pay attention
Australia is not in the beta, but the design’s the bit worth tracking. Once @meta.ai lands here, any user can tag it in a thread that mentions an Australian brand and get a public reply that summarises, contextualises or recommends. For brands, that is a new public channel no one on the comms team can mute on behalf of a customer.
In practice, that’s three jobs for Australian mid-market and enterprise teams. Social listening tooling has to pick up @meta.ai replies as if they were any other public statement. Comms playbooks need a clear plan for when an AI account talks about the brand. And any conversational AI a business already runs inside Threads, whether in DMs or support flows, needs a clean separation from anything @meta.ai posts in feeds.
OAIC guidance on the Privacy Act 1988 still applies to public AI-generated content that contains personal information, even when the author is a platform-owned account. Australian boards should nail down where they stand before the feature lands here, not after.
What this signals for AI on social
Meta is the second of the major social platforms to bake an AI account into the feed, after Grok on X. Google has been pushing AI Overviews inside search, and increasingly inside Discover. For Australian businesses thinking about how they appear in AI answers, this is the same shift moving across surfaces: AI is showing up where users already scroll, instead of waiting for them to come looking.
For AU operators it’s pretty plain. Brand visibility now includes a layer the comms team didn’t write and can’t really opt out of. Treating that layer as part of brand monitoring, not as a moderation edge case, is the cheaper move.
For Australian brands building anything customer-facing in social channels, the real question isn’t whether Meta AI’s beta lands here. It’s whether your comms, social and AI strategy playbooks can handle an AI account that joins conversations you didn’t start.
