Meta’s Muse reached 10.2 million downloads in the US in just 40 days, which makes it among the 12 apps that reached that milestone. However, launching fast does not mean getting sustainable users and revenue, especially because of Meta’s heavy investments in advertisement and new concerns about privacy.A download record bought partly with marketingSince its launch in September, Muse has found its way into the hands of millions of users as it has amassed around 10.9 million downloads amid limited availability in the US and Canada only.Meta is capitalizing on the 3.6 billion daily users across its applications along with aggressive advertising. Sensor Tower says Muse accounted for 39% of Meta’s advertising budget for October compared to only 23% in September. Muse moved up from 73rd to 21st in the list of US advertising-impressions ranking.Muse was ranked higher than ChatGPT, Grok, and Claude as far as downloads are concerned in its initial phase in the US, although ChatGPT still remains the leader globally due to better reach.Retention is the number that mattersEarly engagement looks promising. Muse’s seven-day retention reached 30%, compared with its rivals’ 19% average. After 14 days, the figures were 23% and 16%, respectively. Daily active users reached 2.5 million on October 16, up 15% weekly.Citigroup estimates that by 2030, revenues will reach over $27 billion, which includes roughly $23 billion from transactions and $4.5 billion from subscriptions. The numbers given above are projections and not actual sales.Meta Muse vs ChatGPT, Claude and Grok: Retention Rates and $27B Revenue ForecastAs previously reported by Cryptopolitan, Muse’s launch coincided with the increase of over $200 billion in Meta’s market value. The key question remains how many free users will eventually become paying customers.A sharing flaw undercuts the privacy pitchMuse was designed by Meta with security in mind, providing every single agent with an exclusive and secure virtual machine protected by an additional Sentinel agent that controls internet access.However, Claroc researchers have discovered a flaw in Muse’s privacy functions. They found more than 6,800 publicly discoverable links to shared conversations, some exposing personal budgets, internal financial reports and confidential interview preparations.The breach has affected conversations shared by users rather than private chats. Meta stated on October 20 in a statement to Ars Technica that it is going to suspend sharing, which has turned links that were tested by Claroc inaccessible.Earlier, Reuters revealed that Meta had tested human contractors handling Muse calls. One employee warned:We are one bug away from unnecessary information being leaked to human callers.Meta rolled back that feature.The real prize is agent-driven commerceThe potential for this technology goes beyond Muse. As predicted by McKinsey, AI agents promise to revolutionize marketing by helping individuals choose and purchase their items rather than merely directing them to advertisements.The potential market is enormous. Morgan Stanley estimates that AI agents could influence up to $385 billion in US e-commerce by 2030. Meanwhile, Juniper Research expects annual agent-driven transactions worldwide to reach 120 billion by 2031. These are industry-wide forecasts, not projected revenue for Muse.Muse is already moving into this space through Stripe’s Link checkout, which uses one-time card numbers to protect payment details.For Meta, OpenAI, Google and Anthropic, the competition is no longer just about attracting users. It is also about convincing people to trust AI with their purchases. And every privacy lapse makes that trust harder to earn.The smartest crypto minds already read our newsletter. Want in? Join them.