Meta Launches Personal AI Agent Muse
Left 67%
Center 17%
Right 17%
4 left · 1 center · 1 right
What happened
On Tuesday, Meta launched Muse, a personal AI agent for U.S. users via iOS, Android, muse.ai, and WhatsApp, with a later rollout planned for Meta’s AI glasses. Powered by Meta’s Muse Spark model, it is designed to complete tasks such as sending emails, booking travel, filling out forms, shopping (including checkout via Stripe’s Link), and negotiating on a user’s behalf, and it can keep working after the app is closed, returning when it needs approval. Users choose which apps to connect; Meta says each agent runs on an isolated cloud virtual machine, has no visibility into passwords or payment methods, and is monitored by a separate Sentinel agent, with a confidential encrypted virtual machine planned later this year. Muse is free up to a usage cap, with Power ($20 a month) and Maximum ($100 a month) plans for heavier use; a payment card is required to start, and users must opt out if they do not want interactions used to train Meta’s models.
Sourcing check · Fully on the record · 6 articles
5 of 5 articles cite a named source for the story. It was on the record from the first report.
What these terms mean and how they're counted
- On the record · names a source 5 of 5
- On the record means the article cites a source by name for the story's main claim: a person, or an institution speaking officially, such as a company statement or a named agency's spokesperson. The opposite is an unnamed source. The count is how many of the counted articles cite at least one such source; the names are everyone named across them, each once however the articles spell it. Here: Meta, Alexandr Wang and Mark Zuckerberg
- Primary evidence · cites a document or data 0 of 5
- Primary evidence means the article cites a record you could check yourself: a court filing or ruling, a dataset or report, a recording or transcript, an official statement or press release, market data. A quote is not a document. The count is how many articles cite one. Here: None cited.
- Credits another outlet 0 of 5
- The article says the story comes from another outlet's reporting (“first reported by Reuters”). Those articles relay the story; they are not independent confirmation. Here: Each article presents its own reporting.
- Unnamed sources 0 of 5
- The main claim is attributed to people who are not identified: “people familiar with the matter”, “officials said”. The count is how many articles do this.
- How it developed
- When the first report landed and whether it named anyone; when the first named source and the first document appeared. Here: First report Sep 8, 19:00 UTC by The Verge, naming its source.
- Not counted 1
- Opinion, analysis and video pieces, and articles too short to assess. Every count above is out of the counted articles. An article republished under a new headline is one article.
This is about who is cited, not whether the story is true. Each source below says what it names. How we count
Omitted — what each side leaves out
Unpacked
Left-leaning coverage is more complete on Meta's legal backdrop, which right-leaning coverage omits. CNBC reports Meta just agreed to pay nearly $17 billion to settle state claims it misrepresented harm on Facebook and Instagram, a fact Fox Business never includes. Without that record, a reader cannot see that the company now seeking email, payment, and camera access is still answering for how it handled user harm.
Left-leaning coverage opens on Meta trying to catch up and on public skepticism after a wave of lawsuits. Right-leaning coverage opens on autonomous chores such as booking travel and selling a car, and treats security as an engineering delay rather than a company-history problem.
Who is liable if Muse sends the wrong email or spends money the user did not mean to approve?
Bottom line
A Fox Business reader can watch Muse book travel and never hear that Meta just settled a nearly $17 billion harm case over Facebook and Instagram, the setting CNBC uses, or that Muse chats train models unless you opt out.
The receipts — 6 facts one side's coverage lacks
- Meta agreed to pay nearly $17 billion to settle claims by state attorneys general that it misrepresented harm on Facebook and Instagram
- Users must opt out if they do not want Muse interactions used to train Meta's AI models
- Paid Muse plans are priced at $20 and $100 per month
- Meta delayed Muse's release in April in order to make the product more secure
- Meta is exploring taking a cut of shopping transactions completed by the Muse agent
- A Meta support chatbot helped hackers take over more than 20,000 Instagram accounts
The Left View
Left-leaning outlets frame Muse as Meta’s bid to “catch up in [the] AI race” after costly overhauls, arguing that an agent acting in users’ accounts demands more trust than the company has earned. The Verge says Meta is pitching “the world’s first personal AI agent built for everyone” even though the product is “in fact, very similar” to rivals’ tools, and that “securing trust will be key,” citing Cambridge Analytica, a “grueling trial over the harms of social media,” AI bungles including a support chatbot that helped hijack more than 20,000 Instagram accounts, and a reputation for “pervert glasses.” CNBC notes the debut amid “intensifying scrutiny,” a recent settlement of nearly $17 billion with state attorneys general over alleged misrepresentation of harm on Facebook and Instagram, agent cybersecurity risks, and pressure to monetize AI through subscriptions and a possible cut of shopping. Business Insider highlights Mark Zuckerberg’s low-stakes demos—weekend baking with his toddler, climbing permits, MMA-gym footage—as the public face of a tool those outlets stress still needs mail, payments, and credentials.
The Right View
Fox Business treats Muse chiefly as a capability and markets story, leading with an agent that can “autonomously send emails, book travel on a person's behalf, sell a car and lower a bill,” and describing it as modeled on the open-source agent OpenClaw. It stresses user control—people choose which apps to connect and can revoke access—and Meta’s delay of an April release until it could “cross the threshold” on safety, security, privacy, and performance. The report quotes Vishal Shah, Meta’s VP of AI products, that “every single part of the architecture has been designed to make this as safe, as secure, as private as we can possibly make it,” while noting it is “impossible to say that there is never going to be a mistake.” Coverage situates the launch among other AI-industry headlines and Meta’s share price, without centering past privacy scandals.
Our Take (balanced)
The left’s strongest argument is that Muse only works if users grant Meta reach into email, payments, and other accounts, colliding with a fresh multistate harms settlement and a record of privacy and AI-security failures, so assurances may not win adoption. Right-leaning coverage’s strongest argument is that Meta delayed launch to meet internal safety bars and built isolation, approval gates, and revocable connections specifically for agentic risk, as Shah’s “as safe, as secure, as private as we can possibly make it” claim is meant to show. The central unresolved tension is whether those safeguards and a mass-market, no-learning-curve pitch can overcome a trust deficit when the product’s value requires letting an AI take real-world actions on a user’s behalf.
5 sources
- Meta bets on AI agent Muse to catch up in AI race
- Meta pushes into personal AI agents as company faces public reckoning over privacy and safety
- Mark Zuckerberg laid out what he uses Meta's new AI agent for — kid-friendly cake recipes and MMA advice
- Meta introduces Muse, a personal AI agent that can send emails, book travel
- Meta debuts its Muse AI agent. Will consumers trust it?
More stories
The week's bottom lines, in your inbox
One email a week: the five stories that mattered and what they actually mean. Free.