[DigitalToday reporter Yoonseo Lee] Meta is testing a public reply-style artificial intelligence feature on Threads that explains the context of posts.

On May 12, local time, IT outlet Engadget reported that Meta has started an early beta of the feature through a dedicated Threads account, '@meta.ai'.

The feature works by summoning Meta AI in post replies, similar to xAI's Grok on X, formerly Twitter. Users can talk with the '@meta.ai' account alongside other conversations in the app. They can also summon Meta AI while replying to other people's posts. It is currently in an early beta stage, and the countries where it is available are Malaysia, Saudi Arabia, Mexico, Argentina and Singapore.

Meta gave as an example a question such as, "Why are people talking about the World Cup this month?" Meta AI can respond to such questions in a public reply.

The feature ties in with Meta's expansion strategy centered on its new AI model, 'Muse Spark'. Meta said its recently revamped Meta AI is powered by the model. Meta is also testing 'side chats', a new way of conversing that asks Meta AI for context in group conversations on WhatsApp and other apps. Unlike Threads, this takes place in a separate Meta AI chatroom and is not visible to anyone other than the person who asked.

Muse Spark will also be applied to an in-app version of 'Live AI' in the Meta AI app. It was previously a feature that could be used only on Meta's smart glasses, in which users point the camera at their surroundings and get answers in real time. But criticism has recently been raised that human reviewers could view recordings related to the feature, bringing privacy issues back into focus as Meta expands the use of camera-based AI.

'@meta.ai' is an attempt to bring an AI chatbot into the flow of public conversations on social media, rather than keeping it confined to a separate service. Users can check the AI's explanation directly in the reply area without separately searching for post context or opening another chat window.

The model was introduced as the first result of Meta's effort to build a new foundation for its AI strategy. Meta CEO Mark Zuckerberg (마크 저커버그) has also suggested an AI agent concept similar to OpenClaw based on the model. The public reply-style AI test on Threads is seen as one of the first cases to emerge as Meta moves to place AI across its apps and devices.