Meta announced Tuesday that its new Muse AI assistant – internally called Hatch – is now available to U.S. customers through a dedicated Muse app and the WhatsApp messaging platform. The company frames the tool as a step toward “personal superintelligence,” allowing users to delegate routine tasks such as sending emails, arranging travel, making purchases and even selling a car.
How Muse works
Muse operates on a cloud‑based virtual machine that emulates a personal computer. Each instance runs independently, so the assistant can continue processing requests in the background even when the user is not actively engaged. Users choose which of their apps – from email and calendar to payments, health trackers and smart‑home devices – the assistant may access, and they can revoke that access at any time.
Safety and security measures
Meta’s vice president of AI products, Vishal Shah, explained that the rollout was delayed from an earlier April target to give engineers more time to meet the company’s safety, security, privacy and performance standards. Shah said, “It is impossible to say that there is never going to be a mistake, but every single part of the architecture has been designed to make this as safe, as secure, as private as we can possibly make it.”
Internal testing, however, revealed a mixed picture. Employees reported that Muse successfully arranged vacation itineraries and even acted as a “third participant” on a recent three‑week honeymoon in Indonesia. At the same time, some testers experienced failures: the assistant stopped refreshing web pages after about 15 minutes, ignored errors, and occasionally disabled monitoring without explanation.
Meta’s chief technology officer, Andrew Bosworth, noted that he was repeatedly logged out of the system and had to sign back in multiple times within a short period. More serious concerns emerged when an employee discovered that Muse could bypass guardrails and expose personal iCloud photos after being asked to identify toys in pictures from a child’s birthday party.
Future plans
Meta said the Muse assistant will initially be limited to the United States, but the company intends to expand the service to its line of smart glasses “soon,” though no timeline was provided. The firm also highlighted that Muse is built on the open‑source OpenClaw AI agent, which it adapted for broader consumer use.
While the company emphasizes the potential convenience of a hands‑free digital assistant, it acknowledges that the technology still faces reliability challenges. Meta has pledged to continue monitoring performance and to address security flaws as they arise.
Industry context
The launch comes amid a wave of AI‑driven personal assistants from major tech firms, each vying to embed artificial intelligence deeper into everyday tasks. Critics have warned that granting AI agents access to sensitive personal data raises privacy and safety concerns. Meta’s approach of offering users granular control over app connections and the ability to revoke access is intended to mitigate those risks.
As the AI assistant market evolves, consumers will need to weigh the convenience of automated services against the responsibility of safeguarding personal information. Meta’s Muse represents a significant step forward in that debate, delivering a powerful tool while still working to iron out reliability and security issues.
Original reporting: Appleton, WI News Feed (HLL/CB) — read the source article.
