Almost Everyone is Taking Money from OpenAI. Why is Ziff Davis suing them?

0:26Peter Kafka:We'll see you next time. Millions of items delivered fast. Same day delivery. It's on Prime. Available in select areas. Terms apply.

0:39Introducing Meta Glasses. Hey Meta, any last minute tables for two tonight?

0:45Peter Kafka:Sure, there's a great Italian restaurant 15 minutes away. Hey Meta, where's the nearest flower shop? Five minutes away, straight down Broadway, past the bodega. Their lilies are trending on Instagram. Just saying. Hey, Meta, am I forgetting anything else? How about setting a calendar reminder for next year? Meta Glasses, available in more than 20 styles.

1:13Peter Kafka:From the Vox Media Podcast Network, this is Channels with Peter Kafka. That is me. I'm also chief correspondent at Business Insider. You listen to this podcast so you already know that AI is going to change the way we make, pay for, distribute, and consume media. It's already happening. But just how fast that's going to happen is something we don't know. And we definitely don't know exactly how it's going to happen and what kind of rules of the road, if any, we're going to have as that happens. Which brings us to today's guest, Vivek Shah, the CEO of Ziff Davis. As we discussed, Shah runs a publicly traded company that owns a lot of consumer-focused websites you've heard of, like Mashable or IGN, CNET, and a bunch more you probably don't know.

1:59Peter Kafka:Unlike a lot of other digital media companies, Ziff Davis makes money. It's profitable. Unlike a lot of other digital media publishers, it has not signed a deal with OpenAI, the company that owns ChatGPT. Instead, it's one of only two big publishers, the New York Times is the other, to sue OpenAI. And here's where I should note that Vox Media and National Springer, the two companies I work with and for, have both done open AI deals. So I wanted to talk to Shaw, who got his start in digital media, back at what used to be called Time Inc., about how he thinks AI is going to affect his business. He's actually a bit more optimistic than some other folks.

2:37Peter Kafka:And why he ended up fighting open AI instead of taking their money. I think his answers are pretty candid and useful as we try to sketch out what's coming down the road. So here's me talking to Vivek Shaw. I'm here with Vivek Shah. He is the CEO of Ziff Davis. Welcome back to the show. It's great to be here, Peter. Thanks for having me. I could not remember when we talked before. 2019. Wow. So some things have changed dramatically. Other things, less so. Donald Trump, still president. Media business, still struggling. You guys seem like you're growing, doing well. want to explain to people just I think a bunch of folks will have heard the name Ziff Davis other folks may not know it and may not know what your portfolio looks like these days want to give them the very very brief top line sure I mean I think the simplest description is we buy and grow digital media and internet businesses and we started this about 15 years ago with the acquisition of PC mag which at the time was 50 employees, less than$20 million of revenues.

3:46It came with the corporate names of Davis, which we've obviously kept. And in the intervening 15 years, we've done over 90 acquisitions. We've spent over$3 billion on those acquisitions. And today we generate 1.4 billion of revenues and adjusted EBITDA of$500 million, and we've got 4 ,000 employees. So you're probably not going to find many digital media companies of our size, certainly not with our profits. And we've got a bunch of brands, right? So we can talk about them, but IGN Entertainment and CNET Group and Everyday Health Group and RetailMeNot. And then, you know, more than media, actually, you know, so we've got some software businesses and data businesses.

4:31We own Ookla. We own Viper Security Group. We own Moz, a bunch of stuff, a diversified

4:37Peter Kafka:So when you do, when you check your internet speed and you go to speedtest.com, that's probably going through you. Speedtest.net to be precise, but yes. Sorry. Thank you. Yes. There's some other ones you don't want to, you want to avoid. And the market thinks you guys are worth a little less than$2 billion right now. You'd like, you'd like to increase that, but there's a lot of, there's a lot of pure play digital media companies that would love to be worth$2 billion right now in the, in the public market. So you pulled that off. And I have a perception of you guys as being not vulture buyers, but, but you're often picking up distressed assets.

5:08Peter Kafka:Is that a fair way to think about it? No, I don't think it is. I think we look for opportunities to create value. So, I mean, it might be just helpful. How do we think about M &A? Because it is essentially definitional. It's what we do, right? We're a programmatic acquirer. I think we do a lot of sourcing. We look at a lot of things. So, if we did 90 acquisitions, we've looked at thousands. So, I would say we're selective. I think we're really good at diligencing. I think maybe, you know, that perception comes from the fact that there's a lot of discipline in how we value things. But ultimately, we're looking for businesses where we see a clear path to doing to making the business better.

5:54Right.

5:54Peter Kafka:I think one of the reasons you guys, I have that perception of you is some of the stuff you have bought has been distressed, or at least was worth considerably less than people had valued at previous to that. So Mashable is an example. You guys bid for Gawker Media back in the day when Gawker Media still existed and was coming out of bankruptcy. And so maybe that's why I have that perception of you guys. Yeah. And look, I think in the end, again, Mashable is a great example. Great brand, great business. The fact that it had a large valuation at one point doesn't make it distressed. It just makes it it didn't live up to that valuation, but doesn't make it doesn't mean that it is not a great business because it is.

6:34Peter Kafka:So the last time I talked to you in 2019, we were coming off the era of Facebook fueled growth, both in terms of businesses and valuations and people whose entire pitch was Facebook's going to grow. oh, we're going to go with Facebook. They had tried to, they had to figure something else out. And that is when some of these, some of these media assets were now worth much less than they had been in the past. And there was a lot of concern about the future of media. I feel like we fast forward to 2025 and we're still having that same discussion, except now no one's bothering to complain about Facebook.

7:10Peter Kafka:They're complaining about Google and AI and open AI. And so I want to talk to you about those things. The Google issue is one that seems to affect just about every publisher, right? Even before they've been rolling out their AI answers, there was concern about people's dependency on Google and concern that Google was starting to throttle back traffic. How do you guys think about your relationship with Google and your dependency on it? Well, so look, I think the, you just said it, zero click search, right? This idea that you or in a query and it doesn't result in a click to a website is not new, right?

7:49It first entered the discussion around 10 years ago. And five years ago when we spoke, 65 % of searches at that time did not result in a click. So I don't know what it is today. Let's say it's 70%, even 75%. The reality is that this issue, you know, developed some years ago. And I think in that period of time, I think publishers have been really good at figuring out other platforms, Facebook, Instagram, X, Snap, TikTok, Apple News, Google Discovery, YouTube. I mean, there's Avod, right? There's just a bunch of different platforms. And so, look, I'm not saying search is nothing, but it's not everything.

8:33And I think there's a little bit of, you know, the narrative around is like, you know, the declines in search traffic somehow are existential. And I just don't see it that way. And I don't think it's statistically shown. And so to me.

8:47Peter Kafka:Do you think do you think that's specific to your business that you have a portfolio that is more resistant to changes in Google traffic? Or do you think it's just less of a big deal for the entire industry? You know, that's hard for me to answer. Like, I can't speak to everyone else's businesses and models and what they're doing. I think for ours, we've been very good at adjusting to consumer shifts in consumption behavior. Like that's been something that we've always been focused on. So five, six years ago, I was talking about zero click search. I mean, that was like a narrative. And so now that it's come out so broadly, I kind of feel like, hey, we've been at this for a while.

9:23And so I think if you haven't been, I would accelerate because I don't think this is nothing. But at the same time, I don't think this is new.

9:31Peter Kafka:So, but how do you think about it within your portfolio, right? Because one of the stock answers for, you know, why if a publisher stands up in a public place and says, we feel good about our company. And that's because the stock answer is we have a brand that's meaningful. People are going to come to us. We're a trusted brand. And they're kind of whistling past the graveyard when they say it. But that's the hope. And I can see that working for some of your brands. But I'm assuming a ton of your properties work because people are directed there by probably Google. that they're trying to figure out how to solve some game problem, right?

10:05Peter Kafka:And they type that into some kind of search engine and they get to IGN that way. They're not going to IGN to begin with. So it would seem like a lot of your stuff would encounter that same problem that everyone else has. You know, look, so we've quantified this. And Peter, that's kind of the beauty of being a public company. We have disclosures. So we estimate roughly about 15 % of the company's revenues, which again are$1.4 billion relate to the activity you just described. So again, not nothing. 15 % is not nothing, but it is not 100%. And so you're saying that is intentional. We have intentionally diversified.

10:44Peter Kafka:We didn't just accidentally end up with just 15 % of our business being Google dependent. Correct. Now, as we move to the AI overview world, right, which seems like that will accelerate this idea of Google Zero, how do you think about that? Do you think that the AI answers spell sort of the end of that remaining 15 % of your traffic? I don't. I don't. I think in the end, listen, you know, as humans, we read, we listen, we watch, at least in the context of content and media. And I still think we prefer words and sounds and videos from humans. Do I think that the robots will eat into some of that?

11:27I do. I think the real question is, when do we crave a human voice?

11:32Peter Kafka:Yeah, I'm not talking about AI-created content. But the AI overview is AI-created content. Sure, but you're getting it on Google because you asked for information, and it's giving you the information instead of sending you to the website that has the information. That just seems like if it works correctly, that should negate the reason for having to go to your website. Again, if the robot distillation is sufficient and in cases it will be, then you're right. But I think that if the robot distillation takes away sort of voice and tone and narrative and sources and all the things that I think make human created content great, I think you're going to want to go and dig in deeper.

12:16And so I think the key is just understanding the differences. Look, I look at AI overviews no differently than a decade-long history of changes in the search engine results page. We were talking about this when Google used to have one boxes and all these different ways of providing the answer. And by the way, they contributed to zero-click search. I think this is another one. I don't necessarily think it's just different. So look, I think you have to evolve. I mean, yes, there are going to be queries that are satisfied by the AIO. And the trick, I think, is to produce content where you want to dig deeper, where it's just not a surface answer.

12:58It's just not enough.

12:59Peter Kafka:Why do you think when I talk to your peers, they'll often like, well, I'll ask them about OpenAI. And they sort of refocus the discussion to talk about how upsetting Google is to them. And and I hear murmurs of lawsuits. And in fact, Penske Media, which owns Hollywood Reporter, Variety, Billboard, et cetera, has gone ahead and sued Google. Do you imagine there'll be other folks suing Google over this? You know, it's hard to speculate. I think as I understand, you know, what Penske is looking to do is just try to, it's I think more of a licensing rights dialogue, which is when we give, you know, an entity permission to do one thing, which is to index content and scrape content for search, doesn't necessarily mean we've extended that permission to other use cases, and in this case, AI overviews.

13:51And so I think it actually gets to the heart of all of the issues around intellectual property and copyright and fair use and publishers really asserting their rights.

14:03Peter Kafka:So let's let's segue from that to to the lawsuit you are involved in. You are one of two big publishers suing OpenAI. The other one's The New York Times. You filed a suit a couple of years ago. You filed your suit this spring. just about everybody else, including Vox Media, which helps me produce this podcast, and Business Insider, which pays me as chief correspondent, Axel Springer. They are cutting deals with OpenAI. Why did you sue them instead of cutting a deal? I mean, look, our content's been stolen. We tried in good faith to strike a commercial agreement. you know, OpenAI rebuffed our overtures.

14:43And so, you know, look, we weren't left with, we were left with really the only choice, which was to protect our rights and defend our IP. So

14:51Peter Kafka:we sued. When you say they rebuffed your agreements, I mean, they're cutting deals with lots of folks. So is this sort of just about the economics and you asked for this many dollars and they came back and offered 30 % of that? No, there was no deal offered. So they literally aren't offering you a deal. And so you are suing them. This is pretty similar to the case the Times has filed. They're suing both OpenAI and Microsoft. You're only suing OpenAI. Is it more about the fact that they are taking your work without permission to train the models or is it about them creating work that essentially competes with your work?

15:30So, look, I think the first point you make is the salient point, which is OpenAI has made and continues to make copies, unauthorized copies of our content for training and for retrieval augmented generation. And so that's certainly one of our claims. The other is they're removing our copyright management information in the process and have ignored our robots.txt directives, which that to me is pretty surprising.

16:10Peter Kafka:Yeah, I can get in the weeds on robots and metadata in a second, but just very big picture, right? This law is unsettled, right, about what the LLMs are going to be able to do and not do and how this all shakes out. Right now, the early results seem to indicate that, you know, the courts are fairly accepting of the the AI companies claim that they can go ahead and train themselves on things that are on the Internet. You know, when we've seen cases break down, otherwise, it's because the stuff they're training themselves on was pirated. That's that's that anthropic case. Or it's about the output.

16:51Peter Kafka:Right. You're saying, you know, you don't have the rights to make a Disney Mickey Mouse ripoff just because an engine made it. So it seems like the first part of your argument that they've taken your stuff without permission may not get that far in court. Well, listen, we're going to go through the process and let's see where we come out. We feel like we have very strong arguments. And so I think you just got to you got to play that out. The other thing that I just wanted to point out in this is that there was a time OpenAI would disclose the details of its training. They stopped that a little while ago.

17:28So for GPT-2, which was the last time they actually disclosed details relating to their training, they created something called web text. And the whole concept of web text was only scraping web pages, which were high-quality web pages. And they used essentially signals from Reddit and looking at pages that were linked to from Reddit that had a certain number of karma points. And so that's its way, it was sort of their way of getting at quality. Our pages represented 1 % of that data set, 1 % of the entire training data set. That is not insignificant. And so to me, you know, there's a materiality piece of this, too, because sometimes I think the dialogue is, oh, well, it's the entirety of the Internet.

18:14I think there's a subset of the quality Internet that plays a bigger role in the training of these models.

18:20Peter Kafka:Because that's one of the big debates, right? It's like, well, okay, let's say they did use your stuff without your permission. It does have some value. How much value is there? And the LLM argument is you'd be shocked to learn how little value there is for any particular piece of data, even though it took you a long time to create it, right? We're just ingesting a lot. And your argument is, no, no, you're ingesting quite a bit more of our stuff than other people's. I think that's totally right. And, you know, look, I think if you look right now, so you've got a bunch of things happening, right?

18:50So you've got robots.txt blocking. You've got Citi.

18:53Peter Kafka:Let's stop on the robots.txt. That's code you can put in a web page. It's supposed to tell web crawlers, don't crawl this. It's like a keep out sign. But it's not a law, right? No, it's not. But it's a protocol that's governed the bot use on the web for 30 years. So I think there's this effort to sort of position it as not important. I don't see it like that at all. So you have robots.txt blocking. That hasn't been entirely effective, as you note, because it's been ignored. And so now you have CDN blocking, which at the content delivery network level, which is Cloudflare and Akamai and Fastly, the ability to prevent bots from actually scraping your sites.

19:40And so I think between those activities, the RSL standard, the really simple licensing standard that came out a couple of weeks ago, litigation, I think the one thing that I'll say right now is I'm pretty proud of publishers because they're actually taking action. I think in the past, we've kind of sometimes sat on the sidelines to see how the game's going to play out. That isn't what's happening right now. And I think it's, I think you're seeing it, right? I think there's like a source issue right now with these LLMs. We'll be right back with Vivek Shah, but first a word from a sponsor.

20:20Peter Kafka:In business, the best days are when priorities stay on track. For midsize and large companies, that means managing multiple exposures. from property and liability to cyber and more. The Hartford helps companies plan for risk with insurance and proactive risk solutions. So focus can stay where it belongs. Prepare for risk. Protect success. Smart risk management from The Hartford. Contact your agent or visit thehartford.com slash risk mitigation. And we're live on match day as Doug reaches for a buffalo wing. He's got it. Oh, and he's gone for a can of Pepsi too. What a finish! There's no doubt about it.