Meta Launches Muse AI Agent That Can Send Emails, Book Travel And Make Payments After Instagram Photo Privacy Backlash
Meta has taken a major step towards its vision of AI that does more than simply answer questions. The company has launched Muse, a personal AI agent designed to carry out everyday tasks on behalf of users, including send
Ni Ayush Mukherjee··28K nabasa
Meta Launches Muse AI Agent That Can Send Emails, Book Travel And Make Payments After Instagram Photo Privacy Backlash
Meta has taken a major step towards its vision of AI that does more than simply answer questions. The company has launched Muse, a personal AI agent designed to carry out everyday tasks on behalf of users, including sending emails, booking travel, shopping, managing schedules and even making payments.
The timing is particularly interesting. Meta is launching Muse just two months after another product carrying the Muse name, Muse Image, triggered a privacy backlash over its ability to use images from public Instagram accounts to create AI-generated content. Meta eventually removed that controversial feature, saying it had “missed the mark”.
Now, with its new personal AI agent, Meta is making privacy and security a central part of the pitch. The company says Muse has been built with safeguards designed to keep users in control while allowing the AI to work autonomously in the background.
What Is Meta Muse And What Can It Actually Do?
Muse is designed to behave less like a traditional chatbot and more like a digital assistant that can actually complete tasks. Meta says users can tell Muse what they want to achieve, rather than giving it detailed instructions for every individual step.
For example, a user could ask Muse to plan a trip, and the agent can work through the required steps. It can also handle tasks such as sending emails, filling out forms, shopping, selling an item and organising plans. Meta says Muse can work across apps and services that people use in their everyday lives.
The service is initially available in the US through a dedicated Muse app and WhatsApp. Meta says it is available on iOS, Android and the web, while integration with its smart glasses is planned for the future.
The launch fits directly into CEO Mark Zuckerberg’s long-term AI strategy. Zuckerberg has repeatedly spoken about building “personal superintelligence” that can help people achieve their goals rather than simply generating text or answering questions.
Unlike a conventional chatbot that waits for every new instruction, Muse is designed to continue working on tasks in the background. Meta says it can use a dedicated virtual machine, called Muse Secure VM, which functions like a separate computer with its own browser and computing environment.
Meta Says Muse Has Been Built Around Privacy And Security
This is arguably the most important part of the Muse launch. Giving an AI agent access to email, calendars, shopping accounts, payments and other personal services creates a very different level of risk compared with asking an AI chatbot to write an email.
Meta says users decide which services Muse can access and can revoke those permissions. The company has also created additional security systems designed to monitor the agent’s actions and prevent potentially dangerous behaviour.
Meta’s own security research acknowledges that allowing an AI agent to operate independently created new challenges. The company says its teams have spent significant effort dealing with issues such as prompt injection, long-running tasks, tool use and the possibility of an agent taking an unintended action.
Reuters also reported that internal testing of Muse uncovered security and reliability problems, including concerns around private data and inconsistent performance. Meta delayed the original launch while it worked on these issues and says the current version has reached its minimum safety requirements.
Meta has therefore made a clear promise with Muse: the agent should be powerful enough to do useful work, but users should remain in control of what it can access and what it is allowed to do.
Why The Muse Image Controversy Matters For This Launch
The new Muse launch comes with an uncomfortable piece of recent history.
In July, Meta introduced Muse Image, its first image-generation model from Meta Superintelligence Labs. One of its features allowed people to tag public Instagram accounts and use images from those accounts as references for AI-generated creations.
The problem was that users with public Instagram accounts could be included in the feature unless they opted out. People were not necessarily notified when their publicly available images were used to create AI content. Privacy advocates and others criticised the approach, particularly because of the potential for unwanted manipulation of people’s likenesses.
Meta pulled the feature within days. The company said its intention had been to give people control over how their public content could be referenced, but acknowledged that the feature had “missed the mark”.
That episode makes the privacy messaging around Muse particularly significant. This time, Meta is putting security architecture, permissions and user control at the centre of the product rather than treating them as secondary features.
Muse Could Change How People Use AI Assistants
The bigger significance of Muse is not simply that Meta has launched another chatbot. The company is betting on a future where people tell AI what they want and allow it to figure out how to accomplish it.
That could mean asking an agent to organise a holiday, compare options, fill in information, contact businesses, manage appointments or complete routine online purchases. Instead of producing instructions for the user, the AI becomes responsible for carrying out much of the digital work.
However, that also means mistakes become much more serious. A chatbot giving the wrong travel recommendation is one thing. An autonomous agent booking the wrong flight, sending a sensitive email or making an unwanted purchase is something else entirely.
That is why the success of Muse may ultimately depend less on how intelligent it appears and more on whether users are comfortable trusting it with their digital lives.
Meta Muse Launches Just Before Apple’s Big Event
The timing of the Muse launch is also notable because it arrives just hours before Apple’s major September event. Apple is holding its “Surprise and shine” event on September 9 at 10am PT, which is 10:30pm IST in India.
Apple has already unveiled its next-generation Siri AI as part of Apple Intelligence. The company announced in June that the new Siri would have personal context understanding, onscreen awareness and broader capabilities across Apple’s ecosystem.
However, it would be inaccurate to describe today’s iPhone event as the launch of Siri AI itself. Apple introduced the new Siri at WWDC in June. What happens with Siri and Apple’s wider AI strategy at the September event is therefore something to watch rather than a confirmed new launch.
For Meta, meanwhile, Muse represents a much more aggressive move towards agentic AI: an AI that does not merely talk to you, but acts for you.
And after the company’s recent Muse Image privacy controversy, the biggest question may not be whether Muse can get things done. It is whether users are willing to give an AI agent enough access to actually do them.
Initially limited to users in the US, with access through the Muse app and WhatsApp. Meta says support for additional platforms and smart glasses is planned.