Meta Launches Muse Personal AI Agent; Handle Dispute, Privacy Findings, and Biometric Lawsuit Follow

Meta launched Muse, a US-only AI agent for adults 18+ that runs on a secure cloud VM and can email, shop, book travel, and negotiate via a standalone app and WhatsApp — a rollout shadowed by the band Muse losing its @muse handles, a hands-on review surfacing…

Meta launched Muse on Tuesday, September 8, 2026, its first agentic AI productivity assistant, initially US-only for users 18 and over on iOS and Android via a standalone app and WhatsApp, aligning with Mark Zuckerberg's 'personal superintelligence' vision outlined in a recent 6,500-word essay. Muse runs the agent and user data in a dedicated secure cloud virtual machine and can send emails, browse, fill forms, book travel, shop, and negotiate on users' behalf, asking approval before purchases and sensitive actions with revocable per-app access scopes. Payments run through Stripe Link one-time cards — which Meta calls the first AI agent covered by Link's purchase protection — with Shop Pay and 1Password planned; a second 'Sentinel' agent gates all network access and holds credentials, and a Muse Confidential VM with user-held encryption keys is planned later this year. The Verge's hands-on found Muse deleted thousands of emails and completed an Amazon purchase, but also surfaced detailed personal interests inferred from Instagram and Facebook API data not visible in ad-topic settings; Meta says interactions can be used for AI training but are not shared with its ad systems and can be deleted on request. Sensor Tower puts US iOS downloads at 83,000+ (No. 2 on the App Store, but Android at No. 338 in Productivity), far below Threads' 4.3 million launch-day downloads. The launch drew added scrutiny: Meta took the @muse handles long used by the rock band Muse (trademarked 1999), with execs' mistaken tags of the band called a bug; Meta separately fixed invasive Meta AI suggested prompts about a user's children; and a proposed Chicago federal class action alleges Facebook and Instagram photos were harvested without consent for the unreleased NameTag face-recognition system and to train the Emu and Muse Image models.

• Launched Tuesday, September 8, 2026; US-only, ages 18+, on iOS and Android via a standalone app and WhatsApp; tied to Zuckerberg's superintelligence vision in a 6,500-word essay
• Agent and user data run in a dedicated secure cloud virtual machine; a second 'Sentinel' agent gates all network access and holds credentials; a Muse Confidential VM with user-held encryption keys is planned later this year
• Capabilities include email, browsing, form-filling, travel booking, shopping, negotiation, and media/document generation; approval required before purchases and sensitive actions; per-app access scopes revocable anytime
• Payments via Stripe Link one-time cards, described by Meta as the first AI agent covered by Link's purchase protection; Shop Pay and 1Password integrations planned
• Muse's model reportedly scored 44-48 on the Artificial Analysis Intelligence Index v4.3 (up from 31 for Muse Spark in April), near GPT-5.6 Sol's 47
• Meta says Muse interactions can be used for AI training but are not shared with its ad systems; data can be deleted on request
• The Verge hands-on: Muse deleted thousands of promotional emails and completed an Amazon purchase, but surfaced detailed interests inferred from Instagram/Facebook API data beyond what users see in ad-topic settings
• Sensor Tower: 83,000+ US iOS downloads; climbed from 4th to 2nd on the App Store; Android ranks No. 338 in Productivity on Google Play; comparisons: Threads' 4.3M launch-day US downloads, ChatGPT's 500K first-week installs