Meta AI

So let's jump to our second story here.

Meta now betting on making AI feel safe for the regular person.

On Tuesday, Meta launched Muse, spelled M-U-S-E, a personal AI agent that can shop for you, respond to your emails, book your flights, remind you about your tee times, everything you could possibly imagine on doing all through a conversational app interface like iMessage.

It's Meta's biggest swing yet at turning roughly $130 billion in AI spending this year into something that the average consumer will actually use and help them get a little bit of return on that $130 billion investment.

The pitch from Meta AI's chief, Alexander Wang, was simple.

Quote, "The fact that we've given such a powerful technology in a package that is going to be really digestible for the product is one of the things that will set us apart." Muse is free for most everyday use with a $20 a month and a $100 a month tier for the power users.

Wall Street loved it immediately.

So I think it's really smart.

You have body cam footage, which captured the search and posted online by EWU on their YouTube page.

And you can see Meta AI glasses that were found in the house, along with uniforms bearing the logos of businesses like Chick-fil-A, UPS, Target, Home Depot, Raising Cane's, as well as Amazon.

However, look at this other clip from that very same video.

It's, I guess you can say, a different angle on this whole thing.

You have to remember that Meta isn't just Facebook, Instagram, and Threads.

It's also WhatsApp, Messenger, Meta AI, the Quest headsets, the Ray-Bans, Meta Pay, nine or 10 other products they've acquired or built, and every one of them reports back to the same profile.

If you were to ask me, I'd say they're a data company.

They're a conglomerate of data collecting entities, and each separate entity makes it easier for advertisers to target you.

Contrary to popular belief, Meta glasses are not continuously recording.

They capture your voice after you say, hey, Meta, and photos and videos when you take them.

However, since April of 2025, AF features are on by default, which means Meta AI analyzes the photos and videos you take while those features are on.

And there is a line in the Meta AI terms of service that say by sharing images with Meta AI, you agree that Meta will analyze those images, including facial features, using AI.

With me in the studio today is Andy Halliday and Gareth Hood.

And you are watching and listening to The Daily AI Show.

Some things happened yesterday.

Apple came out, showed us their no-foldy phone.

Two very mixed reviews.

I don't know if folks are excited about that.

I personally, I'm not really excited about it, but the story that we talked about yesterday in terms of Jacob working for OpenAI and then leave Jason

maybe

Beginning this year, we hired a guy named Aminal Dhali.

He was the leader of Meta AI, essentially the Lama models.

We brought him in.

And my ask of him was, I said, Ahmed, I want everybody to become an AI native company.

And I am Matt Duren, LMG's Director of Training.

And today we're going to dig more into the story of OpenAI and how they hacked Hugging Face, a major AI platform.

I think what got scarier is the fact that we recently learned it wasn't just one instance of OpenAI models that escaped.

In fact, 1,200 total models escaped and 700 of them attacked Hugging Face.

Also, what happened to the rest of these models? I mean, there are some missing models there.

And then there are two unnamed companies that they also went after.

I feel like this is a story that is continuing to unravel.

And it also makes me wonder what AI models have hacked other places now that we have yet to find out about.

So the first order is, of course, productivity.

But then the second question is, what does it do to us? How do people communicate with each other? So if we look at meta AI, it has actually connected us superficially, but has made us far more lonely.

Social time in the US, there are some studies, I imagine studies are pretty similar in Europe and other countries.

Social time has decreased by up to 80%.

Yes, I find it very interesting.

AI has greatly contributed to the fact that we don't sit with vulnerability and uncertainty that much anymore.

I forgot the numbers, but an incredible amount of young people have their first romantic experiences with an AI girlfriend, particularly with Meta AI and other agents.

And what does that do? It helps us to just use it as a crutch to not make ourselves vulnerable.

A mother who is a content creator, like she's on this thing, has lots of followers and all that, posted what seemed like a harmless video, singing in the car with her young daughter.

AI followed up by prompting directly beneath the video, it was for meta AI, so either Instagram or Facebook, auto-generated prompts asking who's the child passenger and where does she live? So connecting the dots, when tested, the AI collected the data from different sources across years of activity, including a newborn photo posted on the grandmother's account, previously deleted photos, past location markers to construct a detailed profile of the children accurately deducing their current home location.

So basically, AI discovered exactly how old this little girl was, her address, where she lived, all because AI was scrounging the internet for old pictures, deleted pictures off of various Instagram or Facebook accounts.

So the location is the scary part, right? Stranger danger.