Meta continues refining Meta AI, its own digital assistant based on generative artificial intelligence, to avoid falling behind in the intense AI race in which other giants such as Google and OpenAI are driving mainstream adoption of this technology. Its latest update is Voice Conversations powered by its Muse Spark model.
At the beginning of April, Meta introduced Muse Spark, its new family of LLM models created by its Meta Superintelligence Labs division. The project aimed to deliver a qualitative leap in the company’s AI development, designing Muse Spark as an advanced multimodal AI capable of understanding the user’s visual environment, moving beyond the “text box” approach.
In this way, Muse Spark was integrated into Meta AI, both in the web version and in the mobile app, to improve this digital assistant’s capabilities. Now, Meta has evolved the technology behind this model to allow Meta AI users to hold smooth, natural spoken conversations with the AI while it is working.
This means that, in addition to speaking, users can interrupt the AI, change topics on the fly, or switch from one language to another, without this creating any issues for the system. Meta AI will be able to follow the conversation, maintain understanding, and respond to requests at the same time, whether that involves generating images, showing Reel recommendations, maps, and more.
On top of this, Meta is also adding live artificial intelligence capabilities to the Meta AI app so that people can point it at objects or places in real time and ask questions about them.
Today we’re introducing Meta AI Voice Conversations powered by Muse Spark that let you talk naturally to Meta AI (interrupt, switch topics, or swap languages), and as you talk, Meta AI can generate images and pull up recommendations from Reels, maps, and more. We’re also bringing… pic.twitter.com/97PpSEplgo — Meta Newsroom (@MetaNewsroom) May 12, 2026
Today we’re introducing Meta AI Voice Conversations powered by Muse Spark that let you talk naturally to Meta AI (interrupt, switch topics, or swap languages), and as you talk, Meta AI can generate images and pull up recommendations from Reels, maps, and more. We’re also bringing… pic.twitter.com/97PpSEplgo
— Meta Newsroom (@MetaNewsroom) May 12, 2026
As for the arrival of Muse Spark’s capabilities in the API, the official Meta Newsroom account replied as follows to a user on Twitter: “Muse Spark is available today at http://meta.ai. We are opening a private preview through the API for selected companies, and we plan to open-source future versions of the model later on.”
Another recent development related to Meta AI is its integration into Threads. For now, this rollout is in the testing phase and is only available to users of the social network in Saudi Arabia, Argentina, Malaysia, Mexico, and Singapore.
This new Threads feature allows users to mention Meta AI (@meta.ai) in a thread or reply to ask questions or get context. For example, they could post: “@meta.ai, why is everyone talking so much about the Met Gala this week?” or “@meta.ai, how is Celta de Vigo doing in LaLiga?”
Meta AI’s artificial intelligence will respond publicly through its @meta.ai account using the same language in which the question was asked.
Likewise, if someone wants to reduce the number of Meta AI responses appearing in their Feed, they have several options. First, they can mute the @meta.ai account. In addition, they can select the “Not interested” option on any Meta AI response or choose to hide its replies if they appear directly in their posts.
Photo: Meta and Canva
Your email address will not be published. Required fields are marked *
Δ