Meta's new AI agent "Muse" can shop and book on its own — but raises concerns

In early September, technology company Meta introduced an artificial intelligence agent called Muse, capable of performing practical tasks on a user's behalf — sending emails, booking travel and making online purchases. The company positions it as a personal AI assistant, distinct from ordinary chatbots.

Agents versus chatbots

Unlike traditional chatbots, which mainly respond to user questions in conversation, AI agents can make decisions and act on a user's behalf based on requirements previously established in conversation. This means tools like Muse can independently carry out multi-step tasks rather than simply offering advice or information.

This type of technology has been spreading rapidly across various fields recently — from managing shared work calendars and writing code to planning vacations. AI agents are increasingly being integrated into personal finance management as well — for example, the financial platform Robinhood allows its users to connect third-party AI agents that can make pre-approved purchases automatically.

Reactions and concerns

Following Muse's launch, some users have expressed enthusiasm about the new tool, highlighting its practical usefulness for everyday tasks. However, other users and companies have voiced concern about the spread of such autonomous agents, citing potential privacy and security risks, as well as consequences for other market participants.

One example is Amazon's response — the company blocked Muse from making purchases on its platform, pointing to broader friction among major tech companies over third-party AI agents accessing their systems.

The spread of AI agents also raises questions about how such tools will affect other businesses and their relationships with customers, particularly in situations where agents act autonomously without direct user involvement at every step.