Meta Launches Muse AI Agent That Can Send Emails, Book Travel and Make Payments

Meta has launched Muse, an AI agent designed to perform everyday tasks autonomously, including emails, travel bookings, shopping and payments, while prioritising privacy and security.

Meta Introduces Muse AI Agent for Autonomous Tasks

Meta has expanded its artificial intelligence ambitions with the launch of Muse, a personal AI agent designed to perform tasks for users rather than simply respond to questions.

The new system can assist with activities such as sending emails, planning trips, shopping, filling out forms, managing schedules, selling items and making payments. The idea is to allow users to describe their desired outcome while Muse handles the steps required to complete it.

The launch represents Meta’s broader push towards agentic AI, where artificial intelligence can independently take actions across digital services.

What Is Meta Muse?

Muse is designed to function more like a digital personal assistant than a conventional chatbot. Instead of requiring users to provide detailed instructions for every step, the agent can determine how to complete a requested task.

For instance, users could ask Muse to organise a trip and allow it to work through activities such as researching options and handling bookings.

Meta says Muse can operate across apps and services used in everyday life. The service is initially available in the US through a dedicated application and WhatsApp, with support for iOS, Android and the web.

Meta also plans to bring Muse to its smart glasses in the future.

Muse Uses a Dedicated Secure Virtual Environment

A key part of Muse's architecture is the Muse Secure VM, a dedicated virtual machine that provides the AI agent with its own browser and computing environment.

This setup is intended to allow Muse to perform online activities while keeping its operations separated from the user's primary computing environment.

The technology reflects Meta CEO Mark Zuckerberg's wider vision of developing personal AI systems capable of helping people achieve their goals, rather than simply generating text or answering questions.

Privacy and Security Become Central to Muse

Giving an AI agent access to email accounts, calendars, shopping platforms and payment services creates considerably greater security risks than using a standard chatbot.

Meta says users have control over which services Muse can access and can withdraw permissions when required. The company has also introduced security measures designed to monitor the agent and prevent potentially harmful actions.

Meta's security teams have reportedly focused on challenges including prompt injection, tool usage, long-running tasks and unintended actions.

The company says testing helped identify security and reliability issues involving private information and inconsistent performance. Meta delayed the initial rollout while working on these concerns and says the current version has met its minimum safety requirements.

Previous Muse Image Controversy Raises Privacy Questions

Muse's launch comes shortly after another Meta product called Muse Image faced criticism over privacy concerns.

Muse Image allowed users to reference images from public Instagram accounts while creating AI-generated content. However, concerns emerged because people with public profiles could potentially have their images used without directly being notified.

Privacy advocates criticised the feature over concerns surrounding consent and the potential misuse or manipulation of people's likenesses.

Meta subsequently removed the controversial functionality and acknowledged that it had “missed the mark.”

Because of this recent controversy, Meta's emphasis on permissions and security with the new Muse AI agent is particularly significant.

Muse Could Transform How Consumers Use AI

The biggest change Muse could bring is a shift from AI that provides information to AI that performs actions.

Instead of asking a chatbot how to book a holiday, users could potentially instruct Muse to organise the trip itself. Similarly, the agent could help complete online forms, contact businesses, manage appointments or handle routine purchases.

However, autonomous actions also increase the consequences of mistakes. An incorrect recommendation from a chatbot may be inconvenient, but an AI agent booking the wrong flight, sending confidential information or completing an unwanted transaction could have more serious consequences.

Therefore, user trust and reliable safeguards could become just as important as AI intelligence.

Meta Pushes Ahead in the Agentic AI Race

Muse represents a more ambitious direction for Meta's AI strategy. The company is moving towards systems capable of understanding goals, planning multiple steps and interacting with digital services independently.

The product also arrives amid increasing competition among major technology companies to develop advanced AI assistants that can perform real-world tasks.

For Meta, the challenge will be demonstrating that Muse can combine autonomy, reliability, privacy and user control effectively.

Initial Availability of Meta Muse

Meta Muse is initially restricted to users in the United States. Users can access it through the dedicated Muse application and WhatsApp, with availability across iOS, Android and the web.

Meta has indicated that additional platform support and integration with its smart glasses are planned for the future.

Conclusion

Meta Muse marks a significant step towards agentic AI, where users can delegate digital tasks instead of simply receiving AI-generated answers. Its success will ultimately depend on whether consumers trust the technology with sensitive information, purchases and everyday decisions.