Meta launches Muse, a personal AI agent that acts on users’ behalf

Meta ↗ 6.55% on Tuesday unveiled Muse, a personal AI agent that goes beyond chatbot-style question-and-answer interactions to autonomously handle everyday digital tasks on behalf of users. The agent can send emails, book travel, fill out forms, lower bills, and make purchases — working even after a user closes the app.apnews+1

Muse is available in the U.S. starting Tuesday through a dedicated app on iOS and Android, via the web at muse.ai, and through WhatsApp, with plans to expand to Meta's AI glasses, according to the company's announcement. It is free for basic use, with two paid subscription tiers: Power at $20 per month and Maximum at $100 per month for heavier usage.about.fb+2

How Muse Works

Powered by Meta's Muse Spark model, the agent connects to a user's apps and services — email, calendars, payments, shopping, dining, and more — to carry out tasks with minimal oversight. Users choose which services to connect and how much access Muse receives. When it comes time to pay, Muse uses Link by Stripe to generate one-time-use card numbers, keeping real payment details hidden. Shopify's ↘ 5.45% Shop Pay and 1Password integrations are also planned.about.fb+1

A separate Sentinel agent monitors Muse's activity and gates its internet access, requiring user approval before sensitive actions like sending emails or completing purchases. Meta says Muse has no visibility into passwords or payment methods, and that user conversations and data are not shared with Meta's advertising systems.techcrunch+1

Trust Questions Loom

The launch comes less than two weeks after Meta agreed to an $18 billion multistate settlement over social media consumer harms, as TechCrunch noted. Meta's history of privacy missteps — including a $5 billion FTC settlement in 2019 and the Cambridge Analytica scandal — raises questions about whether consumers will hand over the deeper access an autonomous agent requires.techcrunch

To address those concerns, Meta said Muse runs on a dedicated secure virtual machine called Muse Secure VM, and the company plans to introduce an encrypted "Confidential VM" option later this year where not even Meta can access user data. Users can also opt out of their interactions being used to train Meta's AI models.about.fb

The release places Meta squarely in a crowded field alongside Google's Alphabet Inc. ↘ 2.09% Gemini Spark, Anthropic's Claude Cowork, and OpenAI's agent tools, all competing to define how consumers interact with AI beyond the chatbot era.techcrunch