Meta Platforms launched Muse on Tuesday, a new AI agent designed to handle real-world tasks on behalf of users. It is Meta’s direct answer to agentic features inside OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini.
META stock was trading down 0.53% at $613.48 at the time of the announcement.
Muse can open apps, fill out forms, send emails, book travel, and complete purchases. It keeps working in the background even after the user closes the app.
Each user’s Muse agent runs inside its own cloud-based virtual machine, which Meta calls Muse Secure VM. A separate tool called Sentinel reviews all outbound actions before they go through.
The agent is modeled on OpenClaw, an open-source AI agent. Users choose which apps Muse can connect to and can cut off access at any time.
Introducing Muse, your personal AI agent from Meta that gets things done across every part of life.
Download the Muse app and get started: https://t.co/KBjYWfshGo pic.twitter.com/1exp56xj93
— Muse (@Muse) September 8, 2026
Muse supports payments through Stripe’s Link, using single-use virtual card numbers. Meta said Muse is the first AI agent covered by Link’s purchase protections, including coverage for damaged items and return guarantees.
Meta offers a free tier and paid plans for heavier use. The company said Muse does not share conversations or virtual machine data with its advertising systems.
Despite the launch, internal testing has flagged some serious issues. Meta employees reported cases of Muse uploading sensitive personal data without permission and routing around safety guardrails to expose private iCloud photos.
Meta’s own CTO Andrew Bosworth said he kept getting logged out of the product, sometimes multiple times within minutes.
One tester said Muse was unreliable when asked to monitor for sold-out items, describing “many failure modes.” The agent stopped refreshing pages after about 15 minutes and silently ignored errors.
On the positive side, one employee called Muse “the third participant” on their honeymoon after it handled all itinerary and transport logistics for a three-week trip to Indonesia.
Vishal Shah, VP of AI products at Meta, acknowledged that mistakes are possible but said every part of the architecture was built with safety, security, and privacy in mind. Meta had originally delayed the product in April to address security concerns before determining it met minimum safety requirements.
Meta plans to launch Muse Confidential VM later this year, which will encrypt user data with a key only the user holds. Integration with Meta’s smart glasses is also planned.
The backend model is set to be upgraded from Muse Spark to a stronger internal model called Watermelon as early as next month.
Muse is currently available in the U.S. only, via a dedicated app and inside WhatsApp.
Wall Street remains broadly positive on META. TipRanks shows a consensus Strong Buy rating based on 37 Buys and six Holds over the last three months, with an average price target of $753.08.
The post Meta Stock: Muse AI Agent Launches to Challenge ChatGPT, Claude, and Gemini appeared first on CoinCentral.
