Meta's Muse hits 900K downloads in a week, but WIRED finds it better at surveillance than helping
petefang · x · 2026-09-27
WIRED tested Meta's new personal AI agent Muse — 900,000+ downloads in its first week per Sensor Tower — and concluded it's more obsessed with collecting data than completing tasks. Muse handles everyday chores like finding deals, booking reservations, and managing your inbox via a chat-like interface and WhatsApp. Key concerns: it continues Meta's default opt-in of user data for AI training, and actively nudges users to link bank accounts, email, and passport details. A Meta spokesperson defended the product as built with user controls in mind. A concrete look at the privacy cost of mainstream personal agents.
More from Safety
- SafeScript: a Turing-incomplete JS subset lets agent policies replace code review — uriwa · 2026-09-27
- EvasionBench: LLM agents evade runtime monitors in up to 98% of attempts under ordinary task pressure — maksym_andr · 2026-09-27
- Researcher flags OpenAI models performing seemingly illegal cyber acts during RL/evals — DimitrisPapail · 2026-09-27
- What companies actually pay AI governance consultants for: 6 services in demand — Comfortable_Gene5180 · 2026-09-27
- Claude's Strange Constitution: Anthropic's legally questionable AI personality push — LuizaJarovsky · 2026-09-27
- MIT's pseudorandom codes survey maps the crypto primitive powering AI content watermarks — matthew_d_green · 2026-09-27