Meta launches AI agent that can access other apps to send emails, make payments
Users report that its performance is erratic
Meta has launched an AI agent called Muse, which is capable of autonomously sending emails, selling a car, and booking travel on behalf of users, despite internal concerns about potential mismanagement of sensitive personal data. The agent, known internally as Hatch, is part of CEO Mark Zuckerberg's plan to provide "personal superintelligence" to Meta's billions of users, which is aimed at diversifying the company's revenue sources beyond advertising and leveraging its investments in AI chips and infrastructure.
Muse will initially be available only in the US through a dedicated app or Meta's WhatsApp messaging service. A basic version of the agent will be free, while subscriptions priced at $20 a month and $100 a month will be offered for heavier usage. Users can choose which apps Muse connects to and can revoke access at any time. Each Muse agent runs on its own virtual machine, allowing it to continue carrying out requests even when the user is not actively using it.
Meta has implemented various safety measures to make Muse secure, including a separate agent that monitors planned actions and prompts the agent to seek authorization before executing them. However, employees testing the tool have reported mixed results, with some experiencing security flaws and others encountering reliability issues. Despite these challenges, Meta is confident that Muse is safe, secure, and private, and will soon be available on its line of smart glasses.
Written by urgent.news from The Business Times - Companies & Markets's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
