Meta Disputes Claim That Muse AI Agent Read Private Messages Without Permission

Meta is denying that Muse, its new personal AI agent, read a user’s private messages without permission, turning a fast-growing consumer AI launch into an immediate test of trust and controls. The dispute highlights the central risk for AI agents that can work across email, calendars, messages, browsers and payments: usefulness rises with access, but so does the burden of proving that access is clear, limited and revocable.