What is Muse? The powerful new Meta AI that wants to run your life

Meta released its new "personal AI agent" Muse earlier this month, and the tool has already reached the top of app stores, overtaking ChatGPT, according to The Independent. Meta says Muse "doesn't just answer questions, it actually does the work," taking actions such as sending emails and making purchases on users' behalf, while the company has committed to not using conversations for advertising and to adding encryption controls that only the user can unlock. A New York Times author who tested the system said they were "blown away" after it booked appointments and saved them money, but flagged concerns about data use and safety.

The Independent's journalism is supported by our readers. When you purchase through links on our site, we may earn commission. What is Muse? The powerful new Meta AI that wants to run your life Tool can take action on your behalf, with a view to handling everything automatically - Bookmark Become an Independent member to bookmark this article Want to bookmark your favourite articles and stories to read or reference later? Start your Independent Membership today. Join today https://www.independent.co.uk/subscribe?regSourceMethod=Bookmarks Already a member? Log in Meta https://www.independent.co.uk/topic/meta ’s new AI https://www.independent.co.uk/topic/ai tool, Muse https://www.independent.co.uk/topic/muse , is taking over the world – and it wants to run it. The system was released earlier this month and has already rocketed to the top of app stores https://www.independent.co.uk/tech/meta-ai-agent-muse-chatgpt-b3054937.html , overtaking other competitors to ChatGPT https://www.independent.co.uk/topic/chatgpt . But it promises to do something slightly different to those more traditional chatbots. Meta describes the system as a “personal AI agent”. “It doesn’t just answer questions, it actually does the work,” the company said in its announcement earlier this month, allowing it to do things on its users’ behalf. Muse is built to be used through a chat interface, where users can make requests in natural language, so that it can be used straight away and without technical experience. Meta says that as well as being able to take actions on its owners’ behalf, it will remember details about them and is able to undertake long tasks while users do something else. AI companies have long promoted such “agentic” systems as a potentially transformative use of artificial intelligence. They get the name because they are able to exert agency on their own: taking actions such as sending emails and buying things for their user. But those promises have failed to actually materialise for mainstream users, because of technical limitations as well as concerns about the safety of giving AI systems personal data and letting them act on it. With the introduction of Muse, however, that promise appears to have been at least partly realised. One New York Times author for instance http://go.redirectingat.com/?id=44681X1458326&xcust=b3055240&url=https%3A%2F%2Fwww.nytimes.com%2F2026%2F09%2F22%2Ftechnology%2Fmeta-muse-ai-agent.html&sref=/tech/muse-meta-ai-artificial-intelligence-agent-b3055240.html described how they had given their “life over” to the AI system and found themselves “blown away”, describing how it had booked appointments for them and saved them money. But they also noted that it came with a range of concerns about how data was being used and whether it is safe. After a run of privacy controversies, Meta is clearly aware of the concerns that users might have about giving their data to its system. As such, it has stressed that each version of Muse runs on a devoted computer in the cloud so that each users’ data is securely and separately stored, for instance, and has committed to not using conversations for advertising purposes. Meta has also committed to introducing further privacy controls, including the ability to have the entire virtual computer encrypted with a key that only the user has. That would mean that all of its data cannot be read by Meta. As well as those privacy concerns, agentic AI of this kind is surrounded by fears that it could go rogue and undertake actions that were not asked for or are even outright harmful to a user. Meta has promised that Muse will check in before undertaking sensitive actions such as sending emails or making a purchase, and that it will show a complete trail of everything it has done or has planned. The company “thinks personal superintelligence will be one of the most transformative technologies of a lifetime,” it said in its original announcement. “Muse is a first step: an agent that takes on more of the work so people can focus on what matters to them.” Meta is heavily promoting Muse, with a view to bringing users over from its other products and services. Its mascot is showing in the Instagram app https://x.com/wongmjane/status/2102543064082637085 , for instance, and researchers have found hints that it will soon be available within the company’s smart glasses https://x.com/wongmjane/status/2102737284877672622 . Join our commenting forum Join thought-provoking conversations, follow other Independent readers and see their replies Comments comments-area