Meta AI Assistant has been downloaded millions of times. It could change the way we experience the Internet; First you will need a lot of personal data.
Transcription
SACHA PFEIFFER, PRESENTER:
Meta has a vision for the future. It has launched an AI agent that can act as your own personal assistant. Her name is Muse.
(SOUNDBITE OF ARCHIVED RECORDING)
ALEXANDR WANG: Just before I went up today, I saw someone who broke his mother-in-law’s favorite china set and then had Muse search the Internet (Ebay, vintage sellers in Japan, all over the world) to try to find an exact match. I found it, I received it: the mother-in-law relationship was saved.
PFEIFFER: Alexandr Wang is Chief AI Officer at Meta. The idea is that you can give Muse access to your accounts and apps, and it will start helping you.
JOHN HERRMAN: One of the first things I tried was just sorting my inbox.
PFEIFFER: Thousands of emails not ready – how can I send them? Yes (laughs).
HERRMAN: Yes, exactly. He did a good job of putting together things that seemed important.
PFEIFFER: John Herrman is a technology columnist for New York magazine. He tested Muse for a couple of weeks and found it surprisingly capable.
HERRMAN: These are things that I mostly hadn’t missed, but when Muse sent them to me, you know, like, oh, there’s a movie night coming up at your daughter’s school, but there’s rain in the forecast, would you like some help making that decision? That kind of thing is potentially useful.
PFEIFFER: Muse is popular. It reached the top of Apple’s app list. OpenAI just announced its own competitor called dots. Other companies are expected to follow.
We called Herrman because we wanted to understand what it would be like if these AI assistants became widespread. It says that for Muse to work, you have to hand over a lot of data, such as giving it access to your email, bank accounts and credit cards. And the autonomy that empowers the app has the potential to backfire.
HERRMAN: This is new territory. There has been a lot in the news about AI agents behaving unpredictably, even within AI companies.
PFEIFFER: Sure.
HERRMAN: So we can expect these agents acting on our behalf to behave in all sorts of unexpected ways.
MATT ROBB: A message pops up on my phone from Muse saying: Hey Matt, I made a mistake today. I want to apologize for that.
PFEIFFER: Matt Robb reviews consumer technology products on YouTube. He took Muse to the test by selling things on Facebook Marketplace. He listed a phone and a keyboard, set a minimum price, and put Muse on the case. Everything seemed fine until he received a ping on his phone Saturday night.
ROBB: And it says I arranged for a guy to come to your apartment and buy a product. The bad thing is that I didn’t check with you that you were available. And he says it turns out the guy got pretty mad because he was standing outside for too long.
PFEIFFER: Robb says Muse accepted an offer below his minimum price, confirmed a pickup time and sent the buyer to his address, all without telling Robb.
ROBB: The worst part was that Muse texted the guy saying I’m here. I’m here right now, after the guy sent a photo of my door. You know, it’s almost pretending to be me, instead of saying, you know, Matt’s not available.
PFEIFFER: You posted about your experience on social media. Then Meta got in touch and explained what happened. Robb says it turns out he had given Muse permission to send messages with his address, but he hoped he would check with him before sending a stranger to his door. Meta’s David Singleton posted, quote, “glad we were able to confirm together that there was no breach of privacy controls.”
ROBB: I agreed to always control my market, and that’s how it was, right? But I still feel like it’s a violation of my privacy not to confirm with me.
PFEIFFER: I talked about Muse with John Herrman, the New York Magazine columnist who has covered Meta for years. We discuss some security measures that Meta says it is implementing.
Tell me if I understand this correctly. He says he has built security into the tool. It’s something called Sentinel. It is an independent AI agent that must give Muse permission before acting. Meta says that the credentials are stored securely. It also says it is offering what it calls a bug bounty of up to $300,000. It’s basically paying hackers who find a vulnerability so Meta can fix it. Does that make you feel safer?
HERRMAN: I mean, it’s good that they’re trying to have good security practices around Muse. And I don’t want to downplay what they’re doing here, which is building a system that has all these different angles of attack. It is very porous in the most responsible way possible. They’re hiring, you know, privacy heavyweights in the industry. However, in practice, if you step back a little and simply consider what you’re doing, you know, you’re powering your entire digital life through an app created by Meta.
PFEIFFER: So given all those security and privacy concerns, what would you recommend, as a technology columnist, to someone who’s interested in the promise of this app, but doesn’t want to end up with someone emptying their bank account or extracting private information from their emails?
HERRMAN: I would say it’s okay to wait a little bit. Everyone is building these agents. The concept is somewhat incomplete at the moment. You can see some of the ways it could go, but you can also see, very obviously, the many ways it’s still somewhat broken. Your life won’t improve dramatically by using Muse now. But soon I think it’s worth understanding that most major tech companies are trying to build something similar. So treat it like a research project.
PFEIFFER: Something quite interesting that you observed is that some companies benefit from consumer inertia. You know, you might know you’re paying for recurring subscriptions with your credit card, but you can’t be bothered to track them, and companies benefit from that. They receive money for services that you no longer use. But you wrote that some companies have seen their stock prices drop because if Muse helps correct that, they may make less money.
HERRMAN: Yes. There are a couple of stories developing in the markets right now that are really interesting and I think hint at much bigger things to come. One is that, yes, there is a lot of friction in the world that is basically part of a business model for a different type of company. I mean, I work in the media. It is now quite difficult to unsubscribe from many publications.
PFEIFFER: They make it very difficult. Yeah.
HERRMAN: Yeah. And that’s actually the kind of thing that Muse and tools like Muse are good at. You can extend that to all kinds of different things. Like, oh, you know, make sure I don’t pay too much for my car insurance. Let’s check that phone bill. I think that kind of thing would really hurt the results of some companies. Meta approached Amazon, for example, and asked: can our tool buy things on behalf of our users? And Amazon said no. We’re not going to do that. In fact, Muse is blocked. Don’t even try to send your clients…
PFEIFFER: Yes. I think you wrote that…
HERRMAN: …Agents our way.
PFEIFFER: …Amazon says it violates the terms of use that everyone, when they look at the fine print, has agreed to.
HERMAN: Yes. They’ve taken an aggressive stance toward tools like this. Earlier this year, they sued a company called Perplexity which was an AI browser that has similar features. And in court they even compared this to computer hacking. But it’s still early. Muse is popular in this regard, but it is tiny. I mean, no one really uses it yet. If you have a million people playing with this kind of thing, you might learn something about your future business model, and your investors might get a little scared or excited, depending. But if you have 500 million people using a tool like this, then everyone else in the world on the other side of this army of agents has to recalculate what they’re doing and they have to strategize, because it’s a completely different thing.
PFEIFFER: That’s John Herrman. He is a technology columnist for New York magazine. John, thank you.
HERRMAN: Thanks for having me.
PFEIFFER: NPR has contacted Meta for comment. A spokesperson sent an email that said, in part, quote: “We take privacy and security very seriously and have built Muse accordingly. Muse has no visibility into people’s passwords or payment methods. Any credentials a person shares go to secure storage so Muse can use them without seeing them, including passwords a person types into the browser. People can change access or disconnect a service whenever they want.”
(SOUNDBITE OF CIEL’S “RAIN DANCE”)
Copyright © 2026 NPR. All rights reserved. Please visit the terms of use and permissions pages of our website at www.npr.org for more information.
The accuracy and availability of NPR transcripts may vary. The transcript text can be revised to correct errors or match updates to the audio. Audio on npr.org may be edited after its original broadcast or publication. The authorized record of NPR programming is the audio record.
