Meta launches Muse, a personal AI agent with its own secure cloud VM

Archive entry: this covers an event from September 8, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On September 8, 2026, Meta launched Muse, a consumer personal AI agent powered by Muse Spark, in its newsroom announcement. Muse was rolling out in the US on iOS, Android and muse.ai, and it also worked in WhatsApp.

What changed

• Muse Secure VM: each user's Muse ran in a dedicated cloud virtual machine with its own browser.
• Sentinel: a Sentinel agent gated the agent's actions on the internet.
• Credentials and approvals: credentials were kept in secure storage that the agent could not see. Muse asked before sending email or making purchases and kept a full audit trail.
• Payments: purchases used one-time cards from Link by Stripe. Meta said Muse was the first agent covered by Link's protections, and that support for Shop Pay and 1Password was coming.
• Pricing and platforms: there was a free tier plus paid subscriptions, and support for glasses was coming.
• Muse Confidential VM: planned for later in 2026, encrypted with a key only the user holds.

What it means for people who build with AI

Muse was a mass-market agent that browsed and bought on users' behalf, and Meta described an explicit security architecture around it: an isolated VM, an action gate, hidden credentials and approval prompts. For people building agents, the design showed how a consumer product handled browsing, credentials and payments in one place.

Sources

• Introducing Muse: The World's First Personal AI Agent Built for Everyone (Meta Newsroom, September 8, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.