Muse Goes Mainstream, but Data Fears Follow

Meta’s new AI helper, Muse, is trying to look friendly enough to earn trust. The doll-like agent climbed to the top of the Apple App Store charts after Meta showed it at its Connect event in Menlo Park, California. Meta wants Muse to feel less like a complex computer tool and more like a helpful companion for daily chores, especially for people who are not comfortable with technology.

Meta calls the standard Muse character Jolly. It is cream-colored, with small black eyes and a cheerful smile. The name comes from the idea of a jolly personality, but users can change the agent’s name, look, clothes, and personality. Zuckerberg’s own Muse is named Agrippa and wears a toga with a wreath, giving it a Roman-style look. Meta is using cuteness on purpose, because many tech firms are making AI feel more human through soft looks, petlike voices, and simple personalities.

Agrippa has practical uses too. Zuckerberg said he used it to create a baking recipe and order ingredients for himself and his daughter. He also showed a small keychain-style device, which he called the Muse charm, meant to wake the agent quickly away from a screen. He compared its vibe to a mix of Labubu and Tamagotchi. The key question is whether friendliness and convenience are enough when an agent can sit close to private data.

Muse is also being built into Meta’s AI-powered glasses. Those glasses have faced criticism because some people worry about casual surveillance. Zuckerberg said Meta will keep improving privacy protections for smart glasses. He also said each Muse agent has a private secure computer called Muse Secure VM, where the agent and the user’s personal data are kept. Meta later plans to release Muse Confidential VM, which he says would keep that information hidden even from Meta.

Reactions online were mixed. Wang, Meta’s chief AI officer, posted memes that placed Muse into movies, shows, and historical scenes. Some people followed that lead by making their own avatars and sharing screenshots. Others raised security concerns, with one user asking if Meta really should have access to someone’s whole life. Investor Jason Calacanis said these platforms will need strong defenses and third-party testing. Zuckerberg remains confident in AI despite calls from other tech leaders and government officials to slow down because of serious or existential risks. He argues that market forces should guide AI companies, and that trust plus alignment will decide which agents and models people rely on.