Threads started beta testing a Meta AI integration in May 2026, enabling users to mention @meta.ai in posts or replies for additional context and recommendations, similar to how X's Grok functions [1, 2]. The feature is live in Malaysia, Saudi Arabia, Mexico, Argentina, and Singapore, allowing public account users in these countries to interact with Meta AI directly on Threads [1, 2].
Meta AI replies publicly through the @meta.ai account and responds in the language of the original post to ensure better user engagement and relevance [1, 2]. The AI is designed to provide real-time context about trends, breaking news, and recommendations without users having to leave the Threads platform [1, 2].
Users can control their experience with the AI by muting the @meta.ai account, selecting a 'Not interested' option, or hiding Meta AI responses on their own posts to reduce unwanted interactions [1, 2]. Compared to X's Grok, Meta AI incorporates stronger safeguards, although it remains unclear if it will fully avoid the problems encountered by Grok on X [1, 2].
Meta AI on Threads runs on Meta's latest Muse Spark model, which powers the underlying technology behind the AI responses [2]. Additionally, Meta is testing a different Meta AI feature called "side chats" within WhatsApp group chats. These side chats provide contextual information only visible to the user and operate separately from Threads [2].
The beta test marks the initial rollout of Meta AI integration in social media conversations, with feedback from users in the selected countries expected to guide future development [1, 2].
