We're in our AI slop era

26m
Meta just dropped Vibes and OpenAI just dropped Sora 2. Both social media platforms are filled with AI videos that warp our sense of reality.

This episode was made in collaboration with Vox's Future Perfect. It was produced by Ariana Aspuru, edited by Jolie Myers, fact-checked by Laura Bullard, engineered by Patrick Boyd and Adriene Lilly, and hosted by Sean Rameswaram.

CFOTO/Future Publishing via Getty Images.

Listen to Today, Explained ad-free by becoming a Vox Member: vox.com/members. New Vox members get $20 off their membership right now. Transcript at ⁠vox.com/today-explained-podcast.⁠
Learn more about your ad choices. Visit podcastchoices.com/adchoices

Listen and follow along

Transcript

Today, Explain from Vox, Sean Ramasver.

I'm also from Vox.

I'm soon to be the father of a daughter, and my dad's pretty excited about that because he always wanted a daughter.

He wanted me to be a girl, and that's all the time we have to talk about that.

But the other day, he sent me a video: Hello, Toughy Smile!

A YouTube short of a girl singing.

It was titled 12-year-old sings for Hero Dad, and he said, Hope my granddaughter learns this song one day.

But nothing about this video seemed right.

For starters, the girl looks like she's six and not 12.

The description reads, the heartwarming song that the father taught his daughter before the accident.

What accident?

The mouth movements don't line up.

The shots of the crowd, the lighting, everything is too perfect.

And that's why it had to be AI slop.

I wrote back to my dad, this video is unfortunately artificial intelligence and we haven't talked since.

Just kidding.

But my dad's not the only one getting duped by AI slop.

A lot of you are.

And pretty soon, the rest of us will join you because we are in our AI slop era.

Support for the show today comes from Anthropic, the team behind Claude.

Today's news moves fast, but the most important stories deserve deeper thinking.

Whether you're trying to understand the implications of a policy change or connecting dots across breaking stories, Claude, your new AI collaborator, can help you go beyond the headlines.

Claude doesn't just summarize the news, it helps you explore the context, analyze the patterns, and think through what it all means together.

Try Claude for free at claude.ai/slash today explained.

Adobe Acrobat Studio, so brand new.

Show me all the things PDFs can do.

Do your work with ease and speed.

PDF spaces is all you need.

Do hours of research in an instant.

With key insights from an AI assistant.

Pick a template with a click.

Now your prezzo looks super slick.

Close that deal, yeah, you won.

Do that, doing that, did that, done.

Now you can do that, do that with Acrobat.

Now you can do that, do that with the all-new Acrobat.

It's time to do your best work with the all-new Adobe Acrobat Studio.

You are listening to today.

I'm Hayden Field, senior AI reporter at The Verge.

Slop.

To me, AI slop,

the way I define it is just any form of AI-generated content that's designed to keep you scrolling and keep you consuming and coming back for more.

Today's special is fresh out of the bat.

Little beige, little gray.

We call it AI slop.

Let me get you a scoop.

Which fruit bed would you sleep in?

Blueberry bed.

These new

AI-generated video social media platforms that have an endless scroll or, you know, a AI-generated blog post that goes on and on and on and has a ton of different subsections that aren't needed.

It doesn't feel like there's a lot of intentionality behind it, at least most of the time.

And, you know, a lot of what I see, I would define as slop.

You know, it's like Corgi's running on Hamstead Heath,

like rolling around.

Yeah, it's cute, but guess what?

There's some real videos of corgis out there that I'm sure are just as cute.

You know, it's just kind of like designed to keep you consuming more and more.

And the endless scroll doesn't help with that.

Why is it important that we talk about slop right now?

We've never been in an era with more AI slop.

Everything is AI slop.

AI labs are pivoting to video in a new way.

You You know, we've seen character AI launch an AI-generated video, social media, a scrolling experience.

We've seen Meta do the same thing with this app called Vibes, and now we've seen OpenAI do it with Sora.

It's the most powerful imagination engine ever built.

Sora 2 is also the state of the art for motion, physics IQ, and body mechanics, marking a giant leap forward in realism.

So we're kind of being hit in the face with AI slop everywhere we look.

So it's really important to talk about it right now because, I mean, you're going to be hit with it no matter what.

So you need to know what you're looking at and kind of how to handle it.

Well, let's talk about the two companies really propelling our slop era forward right now, Meta and OpenAI.

Let's start with Meta,

which is really pivoting to slop.

What is Zuck up to?

First of all, I love the term slop era.

That's so true.

We are definitely in our slop era.

It's me.

Yeah, I mean, propelling the slop forward, meta is definitely doing that, maybe more than most other companies.

Vibes, they put it out first before OpenAI put out their app.

To be honest, it was the kind of stuff I would expect to see if Facebook started giving me AI generated videos.

So lots of animal videos,

lots of like blobs being cute, bouncing on the ground.

None of it looked super real to me, and all of it I would definitely define as AI slop.

You once commented fake on your grandma's Facebook post of AI kittens, and you never once thanked Clippy.

The thing that End of the World movies definitely did not prepare me for was having to like enter my Q3 goals and attend all hands meetings.

Let's cast a live-action Candyland movie.

I'll go first.

The obvious choice for Lord Licorice is Snoop Dogg.

It was a lot of stuff that you'd feel like your uncle might send to you and say, oh, look how cute this is.

Or, you know, people of other generations might find more exciting than like Gen Z.

None of it was that convincing.

It was just kind of endless scroll slop designed to keep you coming back for more and be kind of cute and minimally entertaining, in my opinion.

Facebook like changed the world.

Instagram certainly changed our habits.

What is Mark Zuckerberg trying to do with vibes?

So that is the million million-dollar question.

I think a lot of people are asking that online right now.

You know, I saw a lot of anecdotes, reviews, articles, all asking

why did we get this when no one asked for it?

And I think it's honestly, it's all about getting AI into our daily routine.

These companies, especially Meta right now, really want to keep us consuming.

AI generated content and they really want to keep us on the platform.

I mean, we saw that with Facebook.

We saw that with Instagram.

now we're seeing it with vibes.

I think it's really just about Zuckerberg trying to make AI a bigger piece of the everyday person's life and routine and day, getting people more used to it and also kind of putting a signpost in the ground just saying, hey, look, this is where the technology's at right now.

It's a lot better than it was when we saw Will Smith eating spaghetti.

Fresh pasta of Billy.

You know, it's even better than it was when we saw the Pope wearing that puffer jacket.

It's the Pope.

He looks fantastic.

He's wearing a big white puffer jacket that looks like it was made like Balenciaga, Montclair, something like that.

He looks like he just stepped off a runway.

And of course, the kicker is that it's a fake image, that it was made using AI.

It's not a real thing.

So yeah, he's kind of just showing us, hey, the tech has come a long way and it's going to keep going.

How did it get so much better so fast?

Because yes, this is not Will Smith eating spaghetti.

This is is better than that to give Mark and OpenAI, which we'll talk about in a moment, their due.

Is this like the Moore's Law thing where technology is just going to get, you know, exponentially better every so often?

Pretty much, you know, and the other part of it is that you know, AI now trains itself a lot of the time.

It can get better and train itself at getting better.

The only thing really standing in their way, or one of the big things standing in their way, is really just compute.

And, you know, all these companies are building data centers, making new deals every day.

They're really working on getting more computes so that they can push the tech even more forward and faster.

Let's talk about what OpenAI is doing.

They just released something called Sora 2.

What is Sora?

So Sora is their new app,

and it's basically an endless scroll AI-generated video social media app.

So you can think of it as an AI-generated TikTok in a way.

It's honestly a lot like using TikTok.

There's a for you page,

you know, it's a vertical scroll in the same way, but everything is AI generated and it's all using text prompts.

So you can say, can you make me a video of alligators gambling in New York City?

And it'll make it.

Or

yeah, we all need that every single day.

But the craziest part, honestly, is that you can make videos of yourself and your friends too, if they give you permission.

So essentially, it's called a cameo and you record your own face moving side to side.

You record your voice speaking a sequence of numbers and then the technology can parody you doing any number of things that you want.

So you put in a text prompt saying, make a video of me at prom with a bunch of aliens.

And then if your friends give you permission to use their cameos, you can say, hey, make a video of me at prom with a bunch of aliens and also Sean doing Fortnite dances.

It'll create that if you give me permission.

So that's kind of why it's so different than Meta's vibes and why it feels different when you're scrolling through it because you're seeing videos of real people and they look real.

All of the lighting on my face looks really good.

It is just so very, very impressive.

It will never be this bad again.

Let that sink in.

It only gets better.

I was scrolling through and seeing Sam Altman drinking a giant juice box or any number of other things.

Now I'm hanging out with Charizard.

Hey, buddy.

It looks like it's really Sam Altman or it looks like it's really Jake Paul.

Go duck.

The only way I can tell it's not a lot of the time is that they're doing something fantastical or something they just wouldn't be doing.

Have you made one of yourself yet?

I have.

I had to for work.

So I had to for work is a great excuse.

Yep.

I mean, I did want to try it out, but I do sometimes have pause on like giving my likeness to the machine.

But you know what?

I took one for the team.

I did it.

I created a video of myself frolicking through a meadow.

I feel like I could stay forever.

Come on.

Look at these flowers.

They smell amazing.

The sky is incredible.

But I will say,

my voice was not quite right.

And my face was pretty accurate, except for just at the very beginning.

It got me wrong when I was like twirling, but when my face was pretty steady, it looked scarily accurate.

Like some of my friends couldn't tell.

The only things that kind of tipped them off are that my hair was weird, you know, it wasn't a full body scan, things like that.

But yeah, it was incredibly accurate.

And we've come a long way since Will Smith eating spaghetti for sure.

And I imagine your friends aren't like...

80 years old.

They're like your peers.

Right.

That's why it was so terrifying in a way because they couldn't tell a lot of the videos that I showed them were AI.

You know, I have a group chat and the reactions were like, I hate everything about this.

Wait, that's wild.

Why does it look like you?

So people were, yeah, I mean, these are people that are extremely online, just like me, and couldn't really tell.

So it's been, it's been a big change.

Let's talk more about terrifying.

I mean, because it's not not just you posting these in your group chats.

The president has taken to posting AI videos of himself, of his opposition.

Look, guys, there's no way to sugarcoat it.

Nobody likes Democrats anymore.

We have no voters left because of all of our woke trans bullshit.

These aren't just being used for fun.

They're being used to political ends.

How concerned should we be at this point?

I think pretty concerned, especially because a lot of people can't tell tell what's real and what's not.

A lot of the videos that Trump has posted recently that I've seen that have been generated by AI, you could definitely tell they weren't real.

But that doesn't mean the next one he posts won't be realistic.

And of course, it's, you know, October 2025.

There's no election going on right now.

But by the time there is a big federal election, a year and a month from now, this technology will be even better.

Exactly.

And it's interesting because I remember, you you know, back in the day, if you saw a photo, you might say, oh, that's doctored.

But if you saw a video, you'd say, oh, well, this is kind of proof that this thing really happened.

Now we're going to all have to start taking everything with a grain of salt, including videos, because you really just don't know these days what's real and what's not.

I think it's kind of just changing the game on our perception and the lens with which we view the world.

We'll have to take everything with a grain of salt, even more than before, and really just kind of enter this new new era of casting doubt on things unless you have proof.

Kind of how journalists operate, you know, like a lot of times you have to have two or three sources saying the same thing to kind of verify it.

I think the general public will have to operate in a similar way.

Well, how about when we return on Today Explained, you help us and our audience with some tips and tricks to identify AI-generated slop.

Love that.

I need a job with a steady paycheck.

I need a job that offers health care on day one for me and my kids.

I want a job where I can get certified in technical roles, like robotics or software engineering.

In communities across the country, hourly Amazon employees earn an average of over $23 an hour with opportunities to grow their skills and their paycheck by enrolling in free skills training programs and apprenticeships.

Learn more at aboutamazon.com.

Support for today's show comes from Vanta.

If your mind can be blown by compliance automation, let Vanta blow your mind, says Vanta.

Vanta is a trust management platform that says they help businesses automate security and compliance.

Demonstrating trust to customers and prospects is critical to closing deals.

I wouldn't know.

That's according to Vanta.

But it can also be costly and complicated.

Vanta says they help companies of all sizes, large, small, medium, get compliant fast and stay that way with industry-leading AI automation and that continuous monitoring you do know so well.

So whether you're a startup tackling your first SOC 2 or ISO 27001 or an enterprise managing vendor risk, Vanta says their trust management platform makes it quicker, easier, and more scalable.

Vanta also says they help you complete security questionnaires up to five times faster, which could help you win those bigger deals sooner.

You can visit Vanta.com slash explain to sign up for a free demo today.

That's vanta.com slash explained.

Avoiding your unfinished home projects because you're not sure where to start?

Thumbtack knows homes, so you don't have to.

Don't know the difference between matte paint finish and satin?

or what that clunking sound from your dryer is?

With thumbtack, you don't have to be a home pro.

You just have to hire one.

You can hire top-rated pros, see price estimates, and read reviews all on the app.

Download today.

Everything is

AI.

Today, Explain is back.

I'm Sean Romiser.

I'm here with Hayden Field from The Verge, whose friends were astonished in the group chat by how real a video of you looked when you were frolicking in a field.

Is that right?

That's right.

And of course, we started the show hearing about my father, who was fooled by an AI video as well.

So we're going to do a segment here to help all the people who are getting misled by AI video called...

Let's see how this is going to go.

Our producer, Ariana, has thrown a

few videos together, some of them generated by Open AI Sora.

Some might be real.

And you and I are going to collaborate, Hayden, to to figure out which is real and which is not.

And just to be totally transparent here, I have not seen any of these videos.

I have no idea what we are about to see.

Sight.

First up, an empty bottle.

And floats right away.

Now the rock.

Straight down.

You can see the bubbles trailing it.

Bottles still cruising on top.

Cool little company.

Okay, interesting.

So we just saw a guy throwing stuff in what looked to be a pool.

It was scary how realistic those ripples looked and how, you know, the rock did sink.

But I think a lot of of times, when you're trying to figure out if something's AI, you honestly just have to go off the vibes and your intuition because it's so good now that you just can't really tell why you feel something, you just have to go with your gut.

And I will say, the rock in his hand looked kind of weird for a second.

So I'm going to say

AI.

I'm glad you mentioned his hands because I've always known from our episode on the Puffer Pope, which was a while ago, I feel like hands are the hardest thing to recreate, to generate in AI and have them look real.

So, I'm going to join you on Team Slop on this one.

Yay!

Wow, yay!

Thank God.

But I will say, if I saw it quickly while scrolling in like a field of other real things.

Same.

Like, I would never just think it was AI unless I was, you know, being presented with it and saying, oh, is this real or AI?

Especially if I was going quickly.

Okay, so like, you know, no judgments if this would have failed anyone out there.

Ariana, hit us with the second.

Oh, no.

Oh, no.

Okay, this is a video of a cat on a treadmill and it's walking against the motion of the treadmill and then it gets tripped up a few times and runs into the front of the treadmill.

So sad.

That looked pretty real to me.

Yeah, I'm going to say the same thing.

I think

it looked real.

And also, sometimes I feel like if something is too perfect looking, it's AI and it didn't look too perfect.

Like the quality was kind of just normal.

The cat's fur looked normal.

I'm going to say real also.

And I will just say, I will preface this before we get the answer that if I saw this in my feed, I 100%, 100%, 1000% would have just figured it was real.

Yeah, same.

The verdict.

No!

No!

Oh, no!

Oh, my God.

Where are the tail?

Oh.

How is this fake?

Like, look at the cat's whiskers and its fur.

This is crazy.

But you know what?

This is proof that we need to be talking about this because if this fooled us, this is why people need to have really good

judgment and take everything with a grain of salt because, yeah, that looked very real to me.

Okay, well, dad, if you're listening, I'm no better than you are.

I was fooled.

Video three.

You saw him just not use the road on the outside.

A nice sarcastic, I think, was that a thumbs up or a little

Okay, video three was like an F1 race from the perspective of a driver taking a turn and I think just process of elimination one of these has to be real So if it wasn't the cat, it's this one I agree I was gonna say it's real but you know what?

I don't think it looks realer than the cat video, but now I'm gonna say it's real because we only have one left Ariana, was this the real one?

Okay.

Ew.

Okay, two for three, but mostly by process of elimination.

Love it.

Two for three, but like the point of this story is you, an AI reporter,

and I guess I, someone who feels like they're pretty good at telling what's fake and real, apart, both got duped.

Yep, that's really telling.

And so this is why it's so important we're talking about this, because people need to have the tools and, you know, grain of salt in their mind whenever they consume videos like this to not just take it at face value, especially now, you know, people that are extremely online like us, I feel like I always, you know, take pride in the fact that I'm never fooled, but that era is coming to a close, sadly.

Okay, well, Hayden, thank you for playing.

It was a blast.

I mean, is it slop?

I think we're gonna have to be playing that with everyone we know from now on.

Okay, so let's, since we all are playing this game now, whether we want to or not,

help people out there spot red flags or telltale signs that what they're seeing is indeed AI slot?

Because how does one know what they're seeing is real or not in this era where it's getting harder to discern?

As everyone can see, we just got fooled by something.

So these tips I'm about to give you aren't foolproof, but they will help a bit, especially for a video that's a little bit longer.

If you watch something long enough, you'll probably find one of the telltale signs that something's AI generated.

And these are some of them.

One of them is inconsistent lighting.

So it's hard sometimes for AI to get the vibes of a place right.

There's a bunch of lamps, you know, maybe it's really dark in one corner.

Maybe it doesn't have the realistic quality of sunlight.

Basically just inconsistent lighting that isn't quite right.

That could be something you could pick up on.

Another thing is unnatural facial expressions that just don't seem quite right.

So maybe someone's smiling too big or they're crying with their eyes too open, something like that.

Another one is airbrushed skin.

So skin that looks too perfect.

You know, I mean, we all use filters now, but maybe in a video, skin that looks too perfect.

It just doesn't look quite right.

Finally, background details that might disappear or morph as the video goes on.

This is a big one.

Taylor Swift, actually, some of her promo for her new album apparently um had a ferris wheel in the background that uh the spokes kind of blurred as it moved and a coat on two coat hangers merged into one

so yeah i would say watching a video and seeing the background details how they change and really paying attention to the little things okay so there's light there's facial expressions there's airbrush skin there's background details that are a little inconsistent.

Anything else out there that we should be looking for?

AI is still not very good at written words.

I saw like a post-game interview fake video with an athlete, and it was supposed to say Gatorade on the step and repeat behind him, but it said like

instead.

That's my

take on the word.

But it's also interesting because, you know, I just wish we had more rules about this stuff and how it could be disclosed.

You know, for example, OpenAI does have a safeguard in that every video that you download from Sora has a watermark, or at least most videos.

Some pro users can download one without a watermark, but they...

Oh, cool.

So if you pay them money, you could lose the watermark.

Very nice.

But the other thing is I've seen a bunch of YouTube tutorials saying here's how to remove the Sora watermark.

Today, I'll show you how to erase it yourself seamlessly using simple AI workflows.

Let's learn as fast as possible how we remove watermarks in Sora.

I mean, do companies like OpenAI or Meta care if we can tell if this is real or not, or is that exactly what they want?

They say they care.

So I guess that's all we can say right now.

They do have a lot of safeguards in place, at least OpenAI does.

We will make some good decisions and some missteps, but we will take feedback and try to fix the missteps very quickly.

But it's hard because by the very nature of technology like this, it's going to be misused.

So you just kind of have to see if you can stem that misuse as much as possible, which is what they're trying to do, but we're going to have to wait and see how successful they are at that.

And right now, if history is any guide, I'm a little concerned.

Hayden Fields, she reports on AI at theverge.com.

You can also hear her on the Verge's Decoder podcast as of late.

And yet another podcast you might want to check out, Switched on Pop.

They recently did an episode on AI-generated music, and they played a game similar to the one we played today, but with songs.

And in doing so, they inspired the one we played today.

Thanks to game show host Ariana Aspuru for producing today.

Patrick Boyd mixed.

Laura Bullard checked the slop and Jolie Myers edited.

We had an assist from Vox's future perfect.

I'm Sean Ramesfirm with a couple of announcements.

The first, you can support our work right now at vox.com slash members.

We're so eager to have you become a member that we are offering you a deal on membership right now.

If you like deals and memberships or even supporting explanatory journalism, vox.com slash members.

Announcement the second, I am out.

Baby girl is showing up later today, so I'm taking off for five months.

Thank you for listening.

Keep listening.

I'll be back soon.

Happy Halloween, Thanksgiving, New Year's, and Valentine's Day in advance.

Peace.

I need a job with a steady paycheck.

I need a job that offers health care on day one for me and my kids.

I want a job where I can get certified in technical roles, like robotics or software engineering.

In communities across the country, hourly Amazon employees earn an average of over $23 an hour with opportunities to grow their skills and their paycheck by enrolling in free skills training programs and apprenticeships.

Learn more at AboutAmazon.com.

The world is changing faster than ever.

Now, with The Economist Insider, a new premium video offering, we're giving you unprecedented access to the debates shaping our world.

I have sat around that table at NATO.

There is an incoming missile attack now.

Could you answer the question?

I'm sorry, we've got very little time now.

With a few surprises along the way.

I can't promise we'll have a cocktail every time, but we'll try.

So, don't just be an economist reader, get on the inside track with The Economist Insider.

Go to economist.com to join the conversation.