EP 618: RSL vs. the AI Scrape: Can LLM licensing save the open web?

Topics Covered in This Episode:

1. The Open Web Threat from AI Overviews
2. Decline in Website Traffic Due to AI
3. Problems with Current Web Content Licensing
4. Doug Leeds’ Publishing Industry Background
5. Introduction to Really Simple Licensing (RSL)
6. How RSL Web Standard Enables Licensing
7. RSL Collective Blanket License Model
8. Impact of RSL on AI Search Results
9. Big Publishers’ Participation in RSL Collective
10. Benefits for AI Companies Licensing Content
11. Addressing AI Hallucinations with Licensed Data
12. Collective Enforcement Against Content Scraping
13. Legal and Economic Sustainability for Publishers

Keywords:

Really Simple Licensing, RSL, open web, AI licensing, content licensing, online publishing, search engines, AI overviews, large language models, content rights, machine readable licenses, robots.txt, AI web crawlers, content monetization, publisher ecosystem, website traffic, ASCAP model, collective rights organization, AI abstracts, blanket license, RSL Collective, creative commons, content syndication, AI search, web standard, content creators, online media, Google search, paywall content, digital rights management, AI content training, AI citations, legal enforcement, content scraping, AI hallucinations, copyright, content compensation, licensing agreements, AI content quality, legal access to content, user experience, AI compute cost, AI company partnerships, content negotiation, human-written content

Podcast Transcript

Jordan Wilson [00:00:47]:
Is the open web in danger? Sometimes, I think it is. I mean, think of your own use. Are you still going to dozens or hundreds of websites a day, or are you like me? And you're just using, you know, maybe Google's AI overviews or you're using chat GBT or perplexity. So I think part of it is, well, there's great benefits to us humans who can maybe spend less time with 300 browser tabs open, and maybe get better, more accurate answers faster. But what about on the other side? What about all of those clicks and all of those people maybe that aren't going to those websites that are actually feeding these large language models? And this is if you've been listening to the show for a week or a month or a year. This is something I talk about a lot and I care about as a former journalist and as someone that just wants good information, good accurate information in not just today's large language models, but the AI of the future. So, I think one company that we're gonna talk with today is doing something that I've been asking someone to do, so I love that they're doing it. And we're gonna be talking today specifically about real, really simple licensing or RSL and kind of how they might hopefully save the open web.

Jordan Wilson [00:02:05]:
Alright. I'm excited for today's conversation. I hope you are too. If you're new here, welcome. What's going on? My name is Jordan Wilson, and welcome to Everyday AI. This is your daily livestream podcast and free daily newsletter helping everyday business leaders like you and me not just keep up with AI, but how we can make sense of it, get ahead to grow our companies and our careers. If that's what you're trying to do, it starts here with the unedited, unscripted livestream podcast. But to take it to the next level, make sure you go to our website at youreverydayai.com.

Jordan Wilson [00:02:31]:
Sign up for that free daily newsletter. We're gonna be recapping the highlights of today's show as well as all the other AI news from today that you need to know. Alright. Enough chitchat. Let's bring the actual expert on. I'm excited for today's show. So please help me welcome to the stage Doug Leeds, the cofounder of Really Simple Licensing. Doug, thank you so much for joining the Everyday AI Show.

Doug Leeds [00:02:54]:
Thank you, Jordan. Thanks so much for having me.

Jordan Wilson [00:02:56]:
It's great. Alright. So first, let's let's first start. A little bit of your background, right, because you didn't just start this out of left field. Can you tell tell tell everyone a little bit of of your background specifically on the publishing side?

Doug Leeds [00:03:07]:
Yeah. So, well, I started really in in search. I ran a search engine called ask.com or ask jeeves, one of the original search engines. Really, you think back about it. It was trying to do, like, what AI is doing now, but with humans, answering questions instead of sending you the links. But, you know, over the years, Google took more and more of the market share, and started to look, in the early twenty tens for a new business model, and we were investigating where all the traffic was going from search, and it was going to content into media. And given that we had that foothold in search, at least for a little while longer, we started I started to acquire companies, to build a a a media company, that acquired about $600,000,000 worth of, of companies, Investopedia, about.com, dictionary.com, bunch of things like that, rolled them all into one group, under the IC holding company, and, that is now called People Inc because they bought Meredith. And, so it's that thing that we started is now the largest publisher of print and online in in The US.

Jordan Wilson [00:04:20]:
Yeah. So, obviously, if if if you don't know Doug, I know many of you probably will, he's he's top, you know, top of the class when it comes to just publishing on the web. So let's let's just skip straight to it. What is really, really simple licensing or RSL, and what is the problem that you all are trying to solve? Yeah.

Doug Leeds [00:04:40]:
Let me start with the problem because then I can tell you what the the solution is, and it's very simple. Look. AI is great. I use it every single day, and it is a better product than I had when we were when we were giving you search and a bunch of links because it gives you the answer right away. The problem with that is it hurts the ecosystem that the whole web the open web is built on, which is take some of my content, and then index it and then give me a link. So when someone looks for something that I have, you send them to me, and I get the traffic as the publisher, as the creator. Then it's up to me to monetize that traffic, with ads or subscriptions or what have you. But with AI, I just get the answer.

Doug Leeds [00:05:22]:
No need to go to the site. Great user experience, terrible for the ecosystem because without that traffic as a publisher, I can't pay for the content that I'm producing. So So let me let me pause there and make see make sure it makes sense.

Jordan Wilson [00:05:38]:
Yeah. No. That that that makes perfect sense. So, you know, it's something I talk about all the time. Right? Because at a certain point, you need all all of the open Internet to keep publishing, right, if you want AI to be smarter and smarter. You you know, I'm curious. Was what actually led you to say, alright. We need to do something about this.

Jordan Wilson [00:05:58]:
Was there kind of, like, you you know, an epiphany that you had, you know, or or how did you how did you land here with with RSL? Yeah. Well, it my partner,

Doug Leeds [00:06:08]:
by the way, Edgar Walter, who who we worked together twenty years ago at Yahoo, he had created even before that, back in the nineteen nineties at Netscape, he had been one of the cocreators of RSS, which is called really simple, syndication, and it's a a web standard that underpins, you know, a billion websites, I think, at one point. And he came. We were talking. I teach a class at UC Berkeley, and I asked him to come teach talk to the class about his journey as a professional. And at that time, he shared with me what he was working on, which was the updated idea on really simple licensing, which is a machine readable way to just declare your license terms. So very simple, robots.text is a way to say, yes, you can crawl me or no, you can't crawl me, but that's all it can do. It can't do more than that. And we've seen companies like Google say, hey.

Doug Leeds [00:07:02]:
If you wanna be in in search, you have to be in our AI abstracts. You can't say yes to search and no to AI abstracts. It's the same thing. We decided we could do something better, and, really, Eckert was the one who started with the standard saying, look. We can create a open standard that allows anybody to articulate their license terms for any piece of content. We can put that right in robots. Txt, and crawlers can see it, and they can say, oh, I now know what the licensing terms are, and I can then go away if I don't wanna do it, or I can agree to those licensing terms. So that's the standard.

Doug Leeds [00:07:41]:
And then we built a collective, rights organization on top of that. But let me get to that in a second because I wanna make sure I'm covering what

Jordan Wilson [00:07:47]:
you Alright. So, Doug, maybe if there's, you know, a business owner out there, someone that works at a big, you know, media publication. And I I know a lot of, companies maybe a year or two ago just blocked all crawlers when when AI started to come out because they're like, we don't wanna lose traffic. And maybe now they're saying, I might wanna do something else. So tell us, how does the how does RSL work? How do companies enable it? And also, like, what options do they have, you know, to, you know, communicate, to these AI companies, you know, how they want them to crawl or not crawl the site?

Doug Leeds [00:08:21]:
Yep. So, look. Blocking was a technique that made a lot of sense, when you could still get traffic. But, like, things like Google that says you can't block our AI abstracts and be in search means you're basically blocking your ability to show up in search if you do it this way, and simply block. So RSL was created to solve this problem. Rslstandard.org, if you go to the website, you will see, all the the the information about how to use the, the web standard to declare your rights, and those rights can be, put in a machine readable form using the standard language, in your robots. Txt or in your HTTP headers or any piece of content. So anywhere a crawler comes across your content, it will now know what your license terms are.

Doug Leeds [00:09:10]:
And you can put any license terms you want, and there's descriptions of some, like, creative commons ones, up there on the website, but also, you know, pay me or you can use it for this and not for that. All those examples exist on the website, and you can check out and implement it now. But that said, it's not enough to solve the problem Because let's talk about there's millions of websites out there, and each one puts a different license term on, and the AI companies really will base a choice there. Either I I'm not gonna negotiate if I'm the AI company, I'm not gonna negotiate a million times with a million different publishers just because I might want their content. So I'll either steal it, which is what they're doing now, and just take it without paying attention to what your your rule set is, what your license terms are, or I'll just not take it, and I'll go to something else because the web is filled with close competitors of content. Right? So that's why we are building the RSL Collective, which is a collective rights organization. And that's modeled on, like, ASCAP, which has been around for a hundred and ten years in the music business. It basically says, okay, let's take all the content.

Doug Leeds [00:10:26]:
Really, our goal is all the content out there and make it available under a single blanket license that the AI companies can then say, oh, this is the RSL Collective license. I know about that. I will ingest that content. And then they will pay for it, and we will distribute that payment to the content owners. So it becomes really easy, to use. It's in our name, really simple to use and really simple to license, solving problems on both sides.

Jordan Wilson [00:10:53]:
Alright. So I do wanna follow-up

Steven Johnson [00:10:55]:
on the collective side, but before we do quick take a quick thirty second break for a word from our partners. This podcast is supported by Google. Hey, folks. Steven Johnson here, cofounder of NotebookLM. As an author, I've always been obsessed with how software could help organize ideas and make connections. So we built NotebookLM as an AI first tool for anyone trying to make sense of complex information. Upload your documents, and NotebookLM instantly becomes your personal expert, uncovering insights and helping you brainstorm. Try it at notebooklm.google.com.

Steven Johnson [00:11:36]:
This podcast is supported by Google. Hey, folks. Steven Johnson here, cofounder of NotebookLM. As an author, I've always been obsessed with how software been obsessed with how software could help organize ideas and make connections. So we built NotebookLM as an AI first tool for anyone trying to make sense of complex information. Upload your documents and NotebookLM instantly becomes your personal expert, uncovering insights and helping you brainstorm. Try it at notebooklm.google.com.

Jordan Wilson [00:12:10]:
Alright. So, Doug, can you walk us through a little bit on the difference maybe between someone, you know, signing up and implementing RSL and putting it on their website with their own terms versus this collective? Right? Like, are there certain criteria, or, you know, is it paid content only to be a part of this collective? Because that does sound, you know, pretty powerful to be a part of that group, and maybe it also increases your likelihood of ultimately showing up in AI search.

Doug Leeds [00:12:37]:
Yeah. Look. All that is true. So what we are doing right now is we're saying, please go, let us know you exist. Go to the site to the rslcollective.org, different website for the collective, r s l collective dot org, and tell us you exist, and that you're interested in this. We will be releasing a standard license agreement. So r s think of RSL standard as the language, and you can use that to describe anything you want. And then think of RSL Collective as one particular articulation of that language for one particular agreement that you can join.

Doug Leeds [00:13:12]:
And joining is free. It doesn't cost anything to join. You can opt out at any time. It's nonexclusive. But, basically, what you're telling us is, hey. We exist, and please go negotiate on our behalf, for, compensation for our content. And then you will get paid when someone uses the content. So when that shows up in an AI answer, then you get paid, ultimately.

Jordan Wilson [00:13:36]:
So I I I do wanna talk a little bit about some of your your featured supporters. Right? Yeah. When I first looked at this, I'm like, okay. These these are the big names out there. Right? I mean, you have people. You brought up people earlier. You you know, Yahoo, Quora, Ziff Davis, Medium, Stack Overflow, and also Reddit. Right? Because we see we we saw a recent study a couple of weeks ago that said Reddit was actually the most cited or sourced, kinda company out of large language models.

Doug Leeds [00:14:05]:
I saw 40%. Yeah. Which is such such a high amount.

Jordan Wilson [00:14:09]:
Amazing. Like, how might this work? Right? Because we also I've heard that some of these, you know, companies have their own deals. Right? So, like, is is that, like, doubling up or, you know, how does it work with some of these bigger publishers? And does that mean we're gonna see them more in AI search results, less or the same? Yeah. So a a few questions in

Doug Leeds [00:14:29]:
there to unpack. So, so, yes, we have these big companies, and they're they're joining us for two reasons. One, for themselves, which makes sense, because they're businesses, and they're looking out for themselves. But two, for the for the open web. They really are leaders in looking forward and saying, we need an ecosystem that's gonna survive AI, and allow creators of content to get paid. Otherwise, there will be not con there will not be content to train AI on, and there won't be AI. And we all sort of agree that AI is a better product. So how does that work so that these guys, you said they have some of them have separate deals? Absolutely.

Doug Leeds [00:15:11]:
And there's nothing, like I said, nothing exclusive about us. You can go do your own deals. What we believe will happen, once we get this critical mass, which these guys these big guys have helped us get so far, is a better deal with the AI companies. And the reason why is because it's so much easier for them Instead of having to do 20 deals even with the big guys or 50 deals or a 100,000,000 deals, you do one deal, and you pay only when that content is used. Very similar to how Spotify works when they license music rights from ASCAP, or other collective rights organizations, and they pay out when someone plays a song. But if you don't play the song, you don't get paid. And it's up to them to determine what's best use of their product, and

Jordan Wilson [00:15:56]:
they can use the content that makes their product the best, but just pay that's the idea. So you obviously have a very long history of of, you know, being in the Internet publishing, space. I'm not asking you to get out your your your crystal ball here, but it seems like fewer and fewer people are actually visiting websites. Right? So I get how RSL and the RSL Collective can help in general because maybe companies and and publishing companies can still make money without people ever going to their website, you know, if they're getting paid, you know, per crawl, per citation. If there's anything on the agentic side, I think you mentioned that. Right? But what else may still have to happen, right, for even AI models of the future to still have high quality human written content to even train on? Yeah. Well,

Doug Leeds [00:16:48]:
look. First of all, RSL works with AgenTic Web. It works with RAG. It it it works with it it very well works, with MCP, and NL Webb. The the architect of NL Webb was also a cocreator of our standard. So it works in all these formats. It doesn't have to be a a human readable site. So you can put it into DRM content.

Doug Leeds [00:17:11]:
You can put it into, otherwise secure content. So whatever you're publishing, you can put an RSL license on it, and that makes sure that you get paid, by the big AI companies. They want this. Now they may not be saying that, in those words, but they kind of are. They've said, we need a new protocol. Sam Altman said, we need a new protocol so this can happen. In some of the litigation that they're involved in, they've made a statement like, well, the reason we can't license everything is because it's technically impossible. And that, to me, makes me think of the early days of music on the web.

Doug Leeds [00:17:48]:
Napster was taking everybody's music, and the entire music industry was very worried because it was a better product. You didn't have to buy a CD with for $20 and 12 songs you didn't care about. You could make a playlist. You could have just the song that you wanted. But it was not gonna be sustainable in that model because they weren't paying for anything. And to Apple's credit, they went to ASCAP and said, we need a license structure where we can actually license all the music and pay for it. And they created the extension of ASCAP that works for streaming, for digital steering platforms. We're doing the same thing for Internet content.

Doug Leeds [00:18:23]:
It's exactly the same thing. Hey, guys. You're suing. Anthropic just lost you know, settled for 1,500,000,000.0. I'm sure you covered that. Yep. Yep. So that's not the path to go down.

Doug Leeds [00:18:34]:
The path to go down is license everything you want. You build your product the way you think it makes sense for your consumers, but then when you use content, pay for that content just for the ones you've used. That's how it would work.

Jordan Wilson [00:18:47]:
So, obviously, a lot of momentum on the publishing partner side. It seems like you said, right, if if the big tech companies are saying we want something like this, it'll make it easier. Can you share with us maybe a little bit, has there been any reception or momentum on the other side yet? And what might that eventually look like? Yeah. Well, I mean, we're as I

Doug Leeds [00:19:11]:
said, we're a nonprofit, and we are governed by what we call our publisher steering committee. So we are that group is determining what the license terms are going to be. We have received interest, and I can't say more than that. We have received interest from some of the major, LLM companies, but we're putting together what those license terms are with the with the perspective of our our publisher steering committee, because we're nonprofit. So I expect that in not too long, we will have our first deal. But again, if this is interesting to anybody out there, please come

Jordan Wilson [00:19:47]:
to rslcollective.org and just click join, and we'll make sure to include you when we have those deals. Is is if this works out, right, because in in my mind, I'm like, this seems perfect. I've been saying for for multiple years, I'm like, publishing companies have three choices. You can either try to sue the big AI labs, you can partner, or, you know, something now that this exists, the RSL Collective, or you can maybe just go out of business, especially if if you rely on people going to your website, you know, signing up for your email, you know, buying things. Right? But is this if it works out exactly how you're envisioning this, is this something that can both protect the open web, it can protect the humans writing for all of these organizations, and still on

Doug Leeds [00:20:34]:
the other end, guaranteed higher quality, human led training data? Absolutely. And let's talk about that on the other side. So everything you said, I completely agree with. But there's one point that I think, hasn't been covered as much, which is the benefits for the AI companies, not just for licensing, which is great. So they have legal access to this content. But what happens when you have legal access to the content? You can start using it in a way you can't use it now. So they're spending a lot billions of dollars to mash up different sources of content that they've scraped so that they can claim that they're not copying it and reusing it. So what happens when you do that? Billions of dollars in compute cost and subpar answers that have their prone to hallucinations.

Doug Leeds [00:21:19]:
Instead, if your content if they have access to content that actually answers a question, just give that content. And you don't have to mash it up, spend billions of dollars, hurt the planet. You don't have to right? You don't have to, worry about hallucinations. He was like, oh, we have the person perfect article for you. It already exists. It's written by a very knowledgeable human, and here it is because we have the rights to give you that. Mhmm. So the quality is gonna improve when they start licensing this.

Doug Leeds [00:21:46]:
The cost will go down, not just the transaction cost, but the actual compute cost will go down. And so the ecosystem benefits on both sides. Creators get paid, and AI companies get a lower cost of higher quality content. So

Jordan Wilson [00:22:01]:
it's one of those things. Almost sounds too good to be true. Right? Like but what are what are the challenges still? Right? Like, why? Because, again, when you lay this out for me, Doug, I'm like, this sounds great. Sounds like it's gonna solve. It's it's it's a win for publisher, wins for AI companies. In the long run, it's a win for them. Right? And and and it's a win for consumers using all the large language models. Where's where's the short straw? Right? Like like, where's the where's the downside, to all

Doug Leeds [00:22:24]:
of this? I I really don't I think it's we're in the initial phases. We just launched, like, less than two weeks ago, and we're just getting incredible momentum, and it's fantastic. We have to build that group up so that the AI companies will say, this makes sense for us to talk to you about this because we're trying to save our cost of reaching everybody. And if you can do that, that's fantastic. So the big hurdle we have to, go through is getting people signed up. But, again, it's free. It's nonexclusive. And one other point, by the way, you talked about legal enforcement, and it's like sue us if you don't like it has been their approach.

Doug Leeds [00:23:00]:
Well, again, modeled after ASCAP, collective enforcement is part of this. So if you join us and and you start getting infringed or scraped, the whole collective can battle in court for you instead of just you. So this is, again, a model that ASCAP and other collective rights organizations have used for over a hundred years. But instead of just saying, oh, you don't like it? We have billions of dollars, and you have whatever you made last week, lost you in court. You now have the entire industry behind you saying no. If you hurt them, you hurt all of us.

Jordan Wilson [00:23:34]:
Alright. So, Doug, we've been over a a lot on today's conversation. But as we wrap up, what's maybe the one most important takeaway that you think it's important for our audience to hear, when it comes to just licensing content in the future of large language models? What do people need to leave this conversation with?

Doug Leeds [00:23:52]:
I think they need to leave it with hope because that's where I am. The the the LM companies and the the the AI companies are understanding that in order for this to be sustainable, they have to start paying for it, and they need a mechanism to do that in an efficient way, and we provide that. So there's an exciting moment here where we get a better product than we've ever had before. It's more usable, and it supports the people that help create it. Love it. Love it. Can't wait. And, Doug, you know, when when you're ready to announce that first big partnership,

Jordan Wilson [00:24:25]:
you you know, come back and talk with us. We, we we'd love to have you back. But, Doug, thank you so much for taking time out of your day to join the Everyday AI Show. We really appreciate

Doug Leeds [00:24:34]:
it. Thank you. Thank you.

Jordan Wilson [00:24:35]:
If you missed anything, y'all, don't worry. It's all gonna be in our newsletter. A lot of URLs, a lot of information. It's all gonna be in the newsletter. So if you haven't already, please go to youreverydayai.com. Sign up for that free day in the newsletter. We'll see you back tomorrow and everyday for