‘Everyone Is Doing It’: The Truth About AI in Hollywood

If you want a clear understanding of how AI is affecting Hollywood … good luck.

Some of the industry loathes the technology, with screenwriters, directors, and actors alike vowing never to use AI in their work—or collaborate with those who do. Others are much more open to the potential for AI to act as everything from creative partner to cost-cutting engine. And a rare few, Ben Affleck included, are already making millions by capitalizing on AI’s nascent potential to (they argue) streamline and improve the filmmaking process.

At WIRED, we’re constantly doing our best to document AI’s transformational effects on the entertainment business, for better and for worse. (If you haven’t read Ruth Fowler’s recent essay for us, on going from a screenwriter to an underemployed LLM trainer, I highly recommend it.) But in an effort to get the most nuanced possible view, I wanted to talk to someone who knows the industry and its players inside and out.

Matt Belloni, a founding partner at Puck, has covered Hollywood for decades. He writes a scoopy newsletter (What I’m Hearing …) and hosts a scoopy podcast (The Town) that frequently focus on AI and Big Tech as both become increasingly dominant forces across entertainment. Belloni also happens to be a former lawyer, making him the ideal person to ask about the ongoing, and highly consequential, debates over AI and copyright, IP, and celebrity likeness.

Belloni was generous enough to give me an hour of his time, and we managed to cover all the bases—from whether Hollywood types side-eyed Ben Affleck after he sold an AI startup to Netflix for $587 million, to just how much control Big Tech companies now wield over the entertainment industry. You can read our conversation below or listen wherever you get your podcasts.

This interview has been edited for length and clarity.

KATIE DRUMMOND: Matt, I have to start with the obvious, which is artificial intelligence. You've been on top of AI in Hollywood for a very long time, and you've said multiple times that, quote, “everyone is doing it,” meaning everyone [in the industry] is using AI. What exactly do you mean by that?

MATTHEW BELLONI: Well, there's two types of AI, as you know. There's generative AI, which is content creation and things like that, and then there's what I refer to as the corporate AI, like streamlining processes and better search results. Obviously, on the second one, everyone is using it.

I'm not going to name a name, but I heard a very interesting story about a very prominent showrunner who went into the writer's room and basically castigated her writing team for not using ChatGPT to help with the process.

To help with the creative process? Or the admin and bookkeeping?

The creative process. Here's a scenario, suggest some jokes. Here's this, this, this. It doesn't mean they use them. You could do the same thing with Google. It's just better, and people are using it. I know, I've talked to screenwriters who use it to suggest scenarios or to give a description of the inside of a building where they want a scenario to play out, or things that to me seem like part of the process.

You may not know this, but the Writers Guild is actually OK with that. As long as there is a credited writer on every script and that this process doesn't impact jobs, they're OK with the use of this technology.

That's one side. Then the other side is generative and how this is making content. There it's a little murkier, because there are a lot of studios that are purporting to create AI-generated movies, shorts, TV pilots. I've seen a lot of them. They're not great. They're better in animation, and I think this is where we're going to see the disruption first, but you can always kinda tell, and that's the problem when you're trying to relate to people. I can't wait to see what Elon's historically accurate version of The Odyssey looks like.

I mean, it's obviously going to be spectacular. I'm sure he's hard at work on it. He's probably been staying up all night, all week.

First of all, historically accurate is hilarious for a work of fiction. But it just means it's going to be all white people. And the funny thing is, if he tries to use or to scrape the actual version of [Christopher Nolan’s] The Odyssey—I don't know if you saw the headlines, a full version of The Odyssey leaked on X last weekend, and before Universal could send a takedown, it was viewed like a million times …

I actually had not seen that.

So who knows whether Grok got its hands on the actual Odyssey to train for Elon’s fake Odyssey, but whatever.

On the generative AI front, I think where we’re seeing the biggest innovation/disruption is in the animation world, where visually scenes can be completed and storyboards done without the use of artists. You can extrapolate things. Doesn't mean it's the finished product, but even Netflix, on their earnings call this past month, said that 300 shows use generative AI.

I saw that. And they're talking about that relatively loudly and freely. I'm wondering where the stigma is right now. Are there members of Hollywood—actors, directors, screenwriters—who are calling out AI publicly while using it privately? How much of this is still happening behind closed doors in secretive circumstances?

Well, obviously I don't know what everybody does behind closed doors, but I know what people I talk to say to me. I do know that there is still a taint on using it.

A perfect example is Amazon. They did a whole media day where they brought in press and they have this whole AI generative studio that is being run by Albert Chang, one of their executives, and they announced a bunch of shows, including some animation shows, and there were real creators on these shows.

Almost immediately, the backlash that one of the creators got online was so crazy—I don't say crazy meaning bad, I mean crazy like loud—that he stepped down. That's a tangible impact of AI shaming, so to speak, and that is real.

You know, maybe that'll change. Netflix got a lot of leeway on this, because what did they do? They bought Ben Affleck's AI company. They paid almost 600 …

$587 million. And I barely knew he had an AI company until he sold it.

I asked the question, “Was this a $600 million press release?” Because Netflix wants to get into AI. They know that Ben is someone who has a ton of credibility and cachet as a movie star, an Oscar winner, a filmmaker, and having him attached to this company buys them a lot of leeway and allows them to do things that maybe others would bristle at, but it's Ben’s company. He's vouching for it.

And if you look at what that product actually does, it takes the footage that you have shot, and it makes it available for generative AI, essentially.

So it prevents you from having to do a costly reshoot if all you need is some coverage of something. It allows you to extend the periphery of a scene or get a circle shot when you didn't get it in camera. It’s much more of a tool for filmmakers—or at least this is how they've presented it—than something that is replacing [them], because the big fear in Hollywood is that this technology will replace [people].

What was the talk of the town, so to speak, when Ben made that deal? Did people scratch their heads? Did they look at him differently than they had the day before?

I honestly don't think so. There are a number of filmmakers that are doing this, and doing it the so-called right way. Darren Aronofsky is one of them—and Ben, I think because of his credibility and because he's able to articulate very clearly what he's actually doing and what he's not doing.

Because of that, I think that it went over OK. I think the biggest reaction I heard was like, "Holy cow, $600 million?"

That's a lot of money.

He's gonna make more money on that than he's made on any of his movies put together.

Well, he's had some, I don't know, expensive divorces.

I know. The question is, did he launch it when he was married to JLo? I imagine their prenups probably contemplate their separate property.

I think they had a very expensive house they were trying to offload at one point. I'm a big entertainment gossip reader.

There have been a bunch of deals between tech firms and Hollywood. There was the $75 million investment from Google DeepMind into A24, which pissed a lot of fans off. You had Lionsgate taking a stake in Runway, which is a big name in AI-generated video.

Why is this happening? Is it that the technology is ready? Is it that Hollywood is more receptive to this idea that, potentially, the technology can really have an impact on the industry?

I think there's a couple things. First of all, the technology is better, and it is applicable, and there's this fear in the C-suites of the entertainment companies of being left behind. If you talk to people at the studios, one of the biggest problems they face is the cost of movies. The cost of movies keeps going up, whereas the box office is not necessarily covering that.

If you look at the animation space, Pixar is a perfect example. Pixar originals, they can't make them for less than $200 million. I mean, they can, but it's very difficult to make them. They're made in California with a California workforce. They have their own special process that is very labor-intensive, and they haven't used AI to the extent that they maybe could.

If we are looking at a new landscape where movies just aren't going to gross as much as they did, you have to look at bringing the cost down.

And the guilds aren't bringing the cost of labor down. The unions are not willing to do that. So where does it come from? It comes from non-unionized labor that is, for the most part, the below-the-line people and the technology.

That makes me think of Christopher Nolan and The Odyssey. It had a huge cast, it was made all over the world. It's making crazy money. And Chris Nolan, who I believe still professes to not having a phone, not surprisingly has also been critical of AI while also being open to its uses, right?

He's kind of a fence-sitter. But he has also pointed out that younger generations have rejected a lot of quote-unquote slop, right? They're looking for authenticity. That said, are the Chris Nolans of the world going to continue to be even more the exception to the rule?

He is a one of one at this point, where he came into this situation with all the credibility and clout in the world, having won the Oscar for Oppenheimer and having made a billion dollars off that movie. The Odyssey, if it is made at all by another filmmaker, it's made at a much lower cost, not $250, almost $300 million. You don't get to have all those stars in it. You just don't get to do it.

But what I would argue to the studios is that is exactly why it broke through, is that it smells different. If you look at the trailer, if you look at everything around it, it smells authored, and the audience can tell when something is authored.

In a world where you can create all the visual effects you want, there's something about seeing The Odyssey and knowing, because they are very good at marketing this, that this is the vision of one guy who you trust and who is gonna show you something real. And I say real in the sense that movies are increasingly going to be split into movies that are filmed with real people and real scenarios and then those that are just either generative AI in practice or generative AI from the get-go.

There's going to be a split between things that we find authentic and those that we just kind of passively enjoy.

Just sort of consume, for lack of a better term.

That's where social media is going, and the more social video goes that direction, I think the bigger the market opening is for these kinds of big, special, real movies.

But you have a one of one legendary filmmaker like Chris Nolan over here with The Odyssey. Then you have someone like Kane Parsons, who made Backrooms, which made a ton of money for A24. I think it was the highest grossing film that they've ever made, and he actually came out and said he would probably make generative AI disappear if he could.

I'm curious about the degree to which you think Hollywood executives care about a young filmmaker, with nascent influence and nascent clout, having something to say about that. Do they give a shit?

They give a shit in the sense that studios are ultimately in the business of talent relations. They want to appeal to the kinds of people that Kane Parsons represents. The whole reason Christopher Nolan made The Odyssey and Oppenheimer with Universal Pictures was because he was at Warner Bros. for many years, and he woke up one morning in 2020 and all of their movies had been put on HBO Max.

The CEO at the time said, “Because of the situation of the world right now, all of the 2021 movies are gonna be put day and date on HBO Max.” That was an example of the CEO thinking tech forward, “We’re gonna give the consumer what they want. We're going to build our streaming service, even if it means grossing less at the box office.”

And the number one filmmaker in town took his talents down the street. They would love to have the next Chris Nolan movie, but instead, Universal, which didn’t do that, they got the number one filmmaker, and they have an amazing relationship with him.

So I think to the extent that studios care about talent, and A24 in particular very much cares about talent, because they are not only appealing to those filmmakers as a business, they have a brand.

A24 is a brand among people who care about movies, and this Google deal, they took a hit for that. People didn't like it. Their fans did not like it. If that ultimately comes back to bite them, and they sell fewer A24 hats or scented candles or whatever's in their merch store today …

Please tell me they have scented candles.

Oh, they do. I got my podcast producer one of them.

And?

He said it smelled nice.

All right, good. Good for them.

But that's where the studios care. Now, is it gonna stop them? No. They have a fiduciary responsibility to their shareholders to bring down the cost of movies, and they see it as an evolution. You know, if you ever see the movie Babylon—there were a lot of people in Hollywood that were very apprehensive about sound in movies because it wiped out a whole generation of silent movie stars who had terrible voices and couldn't articulate themselves. Brad Pitt in that movie is a perfect example.

Then there were people that didn’t like television. They thought it would kill the business, and the VCR, and digital, and streaming. All the different technological advancements that have upended Hollywood, the studios see AI as another one of those, and that if they manage it correctly, it can lead to more profitability, and, in their mind, more movies and TV shows.

Do you agree with that thesis?

That's the big question. I think it needs to be managed properly, because if they go all in and all of a sudden the NBC fall lineup is a bunch of AI slop …

Right.

I think you lose the differentiator. AI is gonna be everywhere on social media, and you know better than anyone that the tech companies are going all in on it, and you're not gonna be able to distinguish what’s real and what's not real online.

So in the future, if you watch a movie like The Odyssey and you see dolphins coming out of the water, you're gonna think that's real if the studios manage that right. And it was real in that movie. They happened to shoot on a day when dolphins were beside the boat, and they got a great shot of dolphins jumping up out of the water.

And that to me is the differentiator that the Hollywood studios still have, and I would hope that they won’t squander it in pursuit of lowering costs.

I want to get to the labor question in all of this. You talked about animation being a place that's very ripe for this kind of disruption. For animation and everything else in Hollywood, to what extent will unions be able to safeguard the rank and file where AI and job loss is concerned?

We've seen it come up as an issue in labor negotiations twice now, back in ’23 and then this past year as well. [Screen Actors Guild-American Federation of Television and Radio Artists] is not against AI. They will tell you openly, “We are not against the use of AI. We are against the use of AI that causes job loss for our members.”

If you use a synthetic performer in a movie, there is a fee paid that makes it just as costly as using a human actor. So it’s not like, “Let’s populate our movie with AI avatars and save money.” For the Writers Guild, if there is a script on which AI is used, there is going to be a credited writer.

AI cannot be the author of a script for the purposes of a guild-covered movie. And that’s important, because ultimately, what are unions? They are for preserving jobs, and they recognize the technology is out there, but they don’t want it to be a replacement. That’s where it gets to the below-the-line union, and it’s much more difficult, because some people are covered, some jobs are covered, some jobs are not covered.

Some are members of the Teamsters. Some are members of IATSE, which is a below-the-line union for various roles in the business. There are lots of jobs on animated movies that are not union-covered, and those are the ones that I think are the most vulnerable. I’m not prepared to go all the way to what Jeffrey Katzenberg said a few years ago, where he said something like 80 percent or 90 percent of the workforce on an animated movie will be replaced by AI.

I don't think that's gonna be the case, at least not for a while, but they are going to eliminate—they already are. They already are eliminating jobs and automating things.

I wanted to ask you about what some have called an emerging gray market. We did a story about this. We published an essay by a woman who was a screenwriter and started finding work training AI models.

I read that.

I thought it was a very funny piece. I thought it was a little bit sad and dystopian, but this idea was that she is basically training herself out of a job. How much do you hear about that happening, that Hollywood workers who are underemployed for a variety of reasons are actually training AI to make a living, regardless of the stigma attached to the technology?

I mean, to their knowledge or without their knowledge? Because you could argue that every screenwriter is training AI because the tech companies are scraping at will.

We’re gonna get to copyright, but I guess a little bit of both, right?

That was an interesting essay, because I had not heard about that happening. I’m sure it does happen. I’m sure that there are jobs out there where you are essentially inputting different scenarios and inputting different types of conversations in order to train AI. I don’t think that’s super prevalent right now, but on the other side of it, there is a widespread fear, in my opinion, in the screenwriting community that the work of screenwriters is being used to train these models without their consent.

But isn’t that fear totally well-founded? I mean, hasn’t it been established that obviously all of that work, if it’s public, would’ve been ingested by a model?

Yes, but there’s a difference. Well, first of all, I don’t think if you asked Google, OpenAI, Anthropic, on the record, “Do you do this?” they would say yes. Have you actually asked them?

We’ve asked them about how their data sets are trained. I mean, they’re trained on the open web.

Right. But for the most part, most finished movies are not. You’d have to go to piracy.

I see what you mean.

Certainly there are no databases where you can just get every single movie screenplay. Now, there are script databases that you could probably access, but I don’t believe every script that’s ever been written is on the internet.

As we're talking, they must be listening to this and thinking, “What an opportunity for us if we have not yet accessed all of that data.”

Maybe I’m wrong, maybe I’m naive about this stuff. This was a couple years ago, but I asked Aaron Sorkin—I had him for an interview—and I had put a bunch of prompts into ChatGPT and asked it to create a scene of The West Wing set in the Trump White House.

OK.

And it was so bad. It was laughably bad.

They were all sitting instead of walking and talking? What were they doing?

No, they were walking and talking, but it was just generic dialog. It was stuff that didn’t make sense for a character. Like, it just was nonsensical. Maybe that’s improved.

I wanted to ask you about IP and copyright. You used to be an entertainment lawyer, so you are, for many reasons, the perfect person to ask about this. Last year, Disney and Universal sued Midjourney, accusing it of IP infringement because its products could create Shrek or Bart Simpson or Darth Vader. Warner Bros. then joined that lawsuit. Page Six recently called that legal war in its jiujitsu phase, because Midjourney’s …

What does that mean? Is that a fancy word for discovery?

It's a fancy word for “things are getting violent,” maybe. Because Midjourney's lawyers are now demanding that those companies disclose the ways they already use AI.

When you sue someone, you have to know that they’re gonna come for your information, and that’s what you give up when you file a lawsuit: They have access to your documents and communications just like you have access to theirs.

What is your take on that case and on the fair-use issue overall?

So, I have an opinion. But let me first say what the status of the law is, because it’s very unclear. There has not been a definitive determination that using copyrighted material in a model is itself copyright infringement.

The New York Times is in a very expensive legal battle over the same thing.

The better argument for now for these companies to make is that the expression of the output of the model is infringing.

You can take every Disney character and put them into a model, but if what it spits out looks too much like Mickey Mouse, then that itself is infringement, not the ingestion of all of those characters.

Yes, that makes sense.

Now, the entertainment industry would very much like to change that determination and to get courts to rule that the mere act of ingestion is infringement.

We'll see if that ultimately holds water, because it doesn’t look like that's the way things are going. But, you know, there was just this settlement in the book case where a bunch of authors sued, I believe it was OpenAI, it might’ve been Anthropic.

I think it was Anthropic.

It was Anthropic, OK. And the settlement in that one was $1.5 billion.

Yes, it was outrageous.

Yeah, it was not cheap. I think it was the biggest settlement in the history of copyright law, and they’re settling for a reason there. Either the output was too similar or they’re afraid of precedent on the input issue.

It’s going to be fascinating to see how this unfolds over the next five to 10 years, because it’s not clear, and the companies are taking advantage of that vagueness.

Do you have a view here on who should win?

I mean, it’s not going to surprise you that I think that if you put hundreds of millions of dollars into the creation of valuable IP, you should be able to protect that against people who are using it to spit out characters to compete with you.

Sounds entirely reasonable.

Whether that ultimately wins the day, I don’t know, because you have to look at the copyright law framework on this. But to use a fair-use exception to scrape the entire internet, which certainly was not contemplated when the copyright laws were first enacted, I think that’s a stretch.

What about human performers and the idea of voice and likeness? Some are trying to create standards where there aren’t any right now. There's his organization that Cate Blanchett cofounded, RSL Media. The Human Consent Registry is what they have launched, which would be a tool that allows people to control how AI can use their likeness.

Is that the way forward? Is that what we will see more and more of in the future?

I hope so, but that requires buy-in from the various AI companies, and we don't know whether they are going to buy into that. What we have seen is that the entertainment industry is interested in this. CAA, the big talent agency, represents Cate Blanchett. They have thrown their support behind this, and I like that kind of a thing where you have to get consent from a performer or from anyone really to utilize their name, image, and likeness for training models.

SAG-AFTRA has gone in a different direction. They are trying to get a federal right of publicity, because right now this is a hodgepodge of state laws that protect name, image, and likeness.

California has a good one. But if there were a federal right of publicity and an interest in your name, image, and likeness, that would make it easier to protect this federally, and copyright laws are federal laws, obviously.

You've also reported on agencies like CAA and WME meeting with AI companies to basically make deals about their clients’ likeness. What do we know about those deals as they exist now?

Not that much. Do you remember the Tom Cruise–Brad Pitt fight scene?

In what movie?

It was a Chinese model that created a fight scene between Tom Cruise and Brad Pitt.

Oh yes, I remember this. I have an 8-year-old, so I don’t see a lot of movies that don’t involve Moana.

So there was—and this was in the wonderful months when Sora existed …

Oh, yes, an incredible product that so many people wanted.

Especially when you couldn’t use copyrighted images in it. It was really fun when you could make Moana fight Peter Griffin from Family Guy, but then when you couldn't, it was less compelling. So they are trying to have individual meetings with these companies and trying to get a blanket opt-out for all of their clients.

Now obviously, the tech companies, since their inception, have wanted to operate under we will do what we want and you have to opt out, right? The entertainment companies want the default to be opt-out, and if you would like your name, image, and likeness to be included, you can opt in.

That was the big issue with the Zuckerberg AI switch about a month ago. Remember that?

Yes, I do. That actually was a WIRED story.

Yeah, all of a sudden your Instagram photos were gonna be available for anyone to create AI out of.

Unless you knew to go into the settings and tick a button, yeah.

They realized that pissed people off. It’s an ask-forgiveness mentality rather than an ask-permission one.

The entertainment industry would like to change that default. I mean, this goes all the way back to Napster. You have to ask permission, and sometimes permission comes with a fee. So they’re trying to essentially lobby these companies to change the mentality there, and it's an uphill [battle].

I was about to say, there is so much tech money now flooding into Hollywood. Does Hollywood have a shot at this one?

It’s really difficult. I’ll give you a perfect example. The Motion Picture Association is the lobbying group for the studios, and they have a long history. One of their biggest fights over the past 80 years has been piracy, and that’s the big issue they spend all their lobbying and their money on.

So, Sora comes along, right? And you would think this would be a four-alarm fire for the MPA. But guess who’s a member of the MPA now? Amazon.

Of course.

Because they own MGM. So everybody kept asking, “Where’s the statement from the MPA? Where's the statement?” And behind the scenes, Amazon wasn’t going to sign off on an anti-AI-generative-image statement, so they had to go back and forth.

Netflix has mixed feelings about this. We know now that they were talking to Ben Affleck’s company at the time, or were thinking of [it]. So, it’s just a very fraught problem for the industry, because so much tech money is involved. The three biggest buyers of entertainment content right now among the top six are Amazon, Netflix, and Apple.

I have to ask you about Amazon. You have done some phenomenal reporting recently about a Sam Altman film, Artificial, that I am very excited to see but that Amazon MGM suddenly dropped after spending $40 million to make. Tell us more about that mess. How did the movie end up getting made? How did it get dropped, and why?

It's pretty simple. This project was shopped to the market. Amazon was interested in it. It was from a filmmaker, Luca Guadagnino, who they have had a relationship with. He made Challengers for them, and he made a Julia Roberts film last year called After the Hunt for them, and they wanted to do it.

They knew the script. I read the script and wrote about it last summer. At the time I said, “Listen, this is a little dicey, because it’s an indictment of Sam Altman and a scary situation.” It goes into that weekend when the cocreators [of OpenAI] and everybody turned on Sam, and he was kicked out of the company and then essentially got it back and it went on its path …

That was like the Super Bowl for WIRED. That was a wild weekend.

So it essentially dramatizes that event, and nobody looks good except for Ilya. He looks pretty good.

That's really funny.

I've heard from people that that may not …

Yeah, I don't know if that's entirely true, but …

Yeah, but whatever. This is a movie that Amazon wanted to make. They make it, and then all of a sudden in February, it’s announced that they have a $50 billion deal with OpenAI.

They are heavily in bed with that company now. So Mike Hopkins, the head of Prime Video, looks at the finished product. Now, they say the end movie is darker than what they initially envisioned—I haven’t seen the end movie yet; I’ve been told that, yes, it is a dark movie, but that was on the page as well—and they got cold feet.

They essentially ended up paying for someone else to release it. Neon—which is a traditional distributor; they released Anora, the Best Picture winner—they came in, and they got the movie for $0 up front. All they had to do was commit to spending $15 million to release it in theaters, which is what the filmmaker wanted.

Amazon is essentially eating the rest