
Tech • AI • Robotics • Game
Meta’s new AI agent Mus, launched in the United States and Canada on September 8, is rapidly gaining users by automating everyday tasks, but its rise is also intensifying questions about privacy, consumer power and the need for regulation.
Mus moved beyond the chatbot model by connecting to email, calendars and bank accounts to act on a user’s behalf. In the US, it quickly rose to the top of the free App Store rankings on September 18, overtaking ChatGPT, with adoption described as faster than earlier mainstream AI assistants.
The service can book restaurants, buy tickets, analyze subscriptions and handle administrative chores. It can also place calls for users, although some of those calls are still handled through human-operated call centers rather than being fully automated end to end.
After an initial free tier, pricing is presented in a range of roughly $20 to $100 per month. To unlock its most advanced functions, users must grant access to highly sensitive data, including banking information, inboxes and account credentials, making trust and security central to the product’s appeal.
One reported test produced $2,300 in annualized savings in about 90 minutes after the agent identified unused subscriptions, cheaper alternatives and unnecessary software charges. In one example, it spotted a forgotten $49-per-month sports streaming subscription and managed the cancellation process through the user’s browser with approval prompts.
AI agents are also being used to search old accounts for overlooked funds. In one case cited from California, a retired user recovered $730 in unused Amazon credit that had reportedly gone unnoticed for three years.
At Connect on September 23, Mark Zuckerberg made Mus central to Meta’s AI strategy and introduced Mus Charm, a small keychain-style device with a screen and fingerprint sensor. The accessory is designed to run the assistant independently of a phone and is expected for the holiday season, though pricing has not been disclosed.
If AI agents routinely cancel idle services and switch users to better deals, they could undermine business models that benefit from inertia, friction and so-called dark patterns. Annual plans, deeper discounts and simpler retention strategies may become more important as software and media companies adapt to consumers assisted by tireless automated negotiators.
Some analysts have begun warning that AI agents might optimize cash movements between checking and savings accounts so efficiently that they strain traditional banking assumptions. If millions of users automatically sweep spare cash into interest-bearing products, banks could lose part of the stable, low-cost deposits on which many consumer models rely.
Meta’s positioning is notably more optimistic than the apocalyptic tone often associated with leading AI labs. Rather than emphasizing mass job destruction, the company is framing agents as tools that save time, reduce administrative burdens and help households recover money lost to inefficiency.
France does not yet have access to Mus, but the US rollout is offering an early preview of how agent-based AI may enter daily life. The debate is already forming around whether such tools should be deployed quickly and improved through real-world use, or constrained in advance under stricter precautionary rules.
The rollout of Mus suggests AI agents are moving from novelty to utility by taking over small but costly daily tasks. Whether that shift empowers consumers or creates new forms of dependency will depend on how companies, regulators and users handle privacy, errors and accountability.
Ask a question