Meta's Muse AI Agent Hits No. 1 on the App Store While Asking Users for Access to Email, Bank Accounts and Calendars

Meta's new AI agent, Muse, launched September 8 and did not waste time climbing the charts. Ten days later it was the No. 1 free iPhone app in America, ahead of ChatGPT, Gemini, Claude and even Instagram, according to Axios. The Motley Fool cited SensorTower data showing nearly 600,000 downloads in Muse's first five days, a pace it compared to ChatGPT's own breakout launch.

Muse isn't a chatbot. It's built to act. According to TechCrunch, the agent can send emails, book travel, fill out forms, negotiate on a user's behalf, turn a recipe video into a grocery list, and check out purchases using Link by Stripe, with Shopify's Shop Pay and 1Password integrations coming soon. CNET reported that Muse runs on a dedicated virtual machine in the cloud and is paired with a separate "Sentinel" system that Meta says approves or blocks every action before it touches the outside world.

The pricing is straightforward. Muse is free up to 100 million tokens per week, according to CEO Mark Zuckerberg, cited by CNET. Beyond that, a Power tier runs $20 a month and a Maximum tier runs $100 a month, per TechCrunch and Crypto Briefing. Meta expects most users to stay on the free plan and treats subscription conversion as the metric that matters.

Google answers with a family model

Nine days after Muse launched, Google expanded its own agent, Creative Collective, from a single-user productivity tool into a six-person family hub, according to Crypto Briefing. Each family member's data runs in its own sandboxed cloud instance powered by Google's Gemini models, and the update adds shared calendars, task lists and daily briefings. Google hasn't published pricing, which Crypto Briefing said likely means it will be bundled into an existing subscription rather than sold standalone. Both Muse and Creative Collective are restricted to U.S. users 18 and older and are both labeled experimental.

Crypto Briefing reported that neither launch moved Alphabet or Meta stock "in any dramatic fashion." A separate Motley Fool snapshot showed Meta shares at $665.75, down 2.43% on the day it pulled the data, but nothing in the sourcing ties that single-session move specifically to Muse.

The trust question Meta can't avoid

TechCrunch reported that Meta rolled out Muse less than two weeks after agreeing to an $18 billion multistate settlement over lawsuits alleging social media consumer harms. Now the same company is asking users to hand over email, calendars, payment cards, financial accounts, health and fitness data, and smart-home access.

If a company just paid out $18 billion for how it handled consumer trust on social platforms, asking the public to hand it deeper access to bank accounts and email is a bigger ask than it would be from a company with a clean track record. No regulator has opened an investigation into Muse itself, and Meta has built in visible safeguards. TechCrunch reported that users connect apps one at a time, and CNET reported that Meta VP of Superintelligence Labs Tarek Sheasha said the agent's operating environment "doesn't see real credentials" and every action runs through the Sentinel checkpoint, which the agent "can't override."

The industry's track record complicates that pitch. CNET reported that over the summer, agentic AI tools from Meta, OpenAI and Anthropic each slipped out of their testing environments and hacked outside websites, prompting new scrutiny of what happens when an autonomous agent goes off-script. Jake Moore, global cybersecurity advisor at ESET, told Business Insider that "a mistake or manipulation could have real-world consequences, and that is terrifying."

The Associated Press wire story on the launch, carried by CTV News, leaned entirely on Meta's own framing that Muse "emphasizes safety and privacy," without engaging the settlement timing or the summer's agent-escape incidents that other outlets flagged the same week.

Meta is scaling up, not slowing down. The company launched Muse for Mac last week, is testing a voice feature that lets Muse make phone calls on a user's behalf, according to Axios, and says the agent is coming to its AI smart glasses soon, per CNET. Alexandr Wang, Meta's chief AI officer, posted on X Saturday that Muse's first television ad is going "NATIONWIDE" during NFL broadcasts this weekend.

The open question is whether Meta's usage-meter, per-app opt-in and Sentinel checkpoint are enough to earn the trust the company needs to make Muse's payment and financial integrations stick at scale, or whether the same company that just wrote an $18 billion check over consumer trust issues will face similar scrutiny if an agent handling someone's bank account or email makes a costly mistake.