LIVE
+++ Anthropic CEO Dario Amodei calls for a deliberate AI slowdown — Altman and Musk agree, Trump rejects it +++ OpenAI, Anthropic, and Google DeepMind have coordinated on AI safety for weeks, confirms OpenAI policy chief Chris Lehane +++ Apple opens the public beta of its new Gemini-powered Siri — for now without several EU countries and China +++ Meta launches personal AI agent Muse for 20 to 100 dollars a month — its own staff already flag security flaws +++ Nvidia's upcoming Rubin platform reaches up to 67 times the efficiency of the current Blackwell generation, per SemiAnalysis +++ Intel spin-off Cornelis Networks raises 205 million dollars and partners with Qualcomm on new AI networking technology ++++++ Anthropic CEO Dario Amodei calls for a deliberate AI slowdown — Altman and Musk agree, Trump rejects it +++ OpenAI, Anthropic, and Google DeepMind have coordinated on AI safety for weeks, confirms OpenAI policy chief Chris Lehane +++ Apple opens the public beta of its new Gemini-powered Siri — for now without several EU countries and China +++ Meta launches personal AI agent Muse for 20 to 100 dollars a month — its own staff already flag security flaws +++ Nvidia's upcoming Rubin platform reaches up to 67 times the efficiency of the current Blackwell generation, per SemiAnalysis +++ Intel spin-off Cornelis Networks raises 205 million dollars and partners with Qualcomm on new AI networking technology +++
All news ›
AI IN LIFE AI IN LIFENEWS
DAILY
AI Agents

Meta launches Muse: a personal AI agent that carries out tasks on its own

Muse books travel, writes emails, and shops — for a fee. Meta's own staff have already flagged security flaws.

Meta launches Muse: a personal AI agent that carries out tasks on its own

Illustration · AI-generated (AI IN LIFE)

At a glance

  • Launch: September 8, 2026, US only for now, on iOS/Android/web (muse.ai)
  • Engine: "Muse Spark 1.3" model, running on "Muse Secure VM"
  • Pricing: free base tier plus $20 and $100 monthly plans
  • Approval requirement: users must explicitly approve sensitive actions (messages, spending)
  • Security concern: Meta's own staff reportedly already flagged security flaws, per Forbes

Meta unveiled Muse, a personal AI agent, on September 8, 2026. Instead of merely answering questions, it carries out tasks on its own. The product advances Mark Zuckerberg's vision of giving every user a personal AI assistant.

Rather than just responding to queries, Muse takes actions on the user's behalf: sending emails, booking travel, filling out forms, and shopping online. The rollout begins in the US on iOS, Android, and the web at muse.ai, with support for Meta's AI glasses coming soon.

Under the hood, Muse runs on a dedicated, secured virtual machine called Muse Secure VM and is powered by Meta's most capable model to date, Muse Spark 1.3. The service works both as its own app and inside WhatsApp.

Pricing includes a free tier that Meta says covers most everyday needs, plus two paid plans at 20 and 100 dollars a month for heavier use. Meta says every sensitive action — such as sending a message in the user's name or spending money — requires explicit user approval first.

Shortly after launch, Forbes reported that Meta's own staff had already flagged security flaws in the new agent. A more locked-down mode called Muse Confidential VM is planned for later in 2026.

◈ AI-GENERATED REPORT · SOURCES LINKED

FAQ

How is Muse different from a regular chatbot?

A regular chatbot answers questions. Muse is designed to actually act on the user's behalf — booking a trip or submitting a form, for example — rather than just giving instructions for how to do it.

Is it safe to let Muse handle payments or messages?

Meta says it requires explicit user approval before every sensitive action. Reports indicate, however, that Meta's own staff have already flagged security flaws, so caution is advised until those are resolved.

Is Muse coming to Europe or Germany?

The launch is US-only for now. Meta has not announced a timeline for international expansion.