HTTP 200 OK
Allow: GET, PUT, PATCH, DELETE, HEAD, OPTIONS
Content-Type: application/json
Vary: Accept
{
"resource_uri": "https://www.courtlistener.com/api/rest/v4/audio/105475/?format=api",
"id": 105475,
"absolute_url": "/audio/105475/thomson-reuters-enterprise-centre-gmbh-v-ross-intelligence-inc/",
"panel": [],
"docket": "https://www.courtlistener.com/api/rest/v4/dockets/70622297/?format=api",
"date_created": "2026-06-11T14:32:14.607483-07:00",
"date_modified": "2026-06-11T14:54:44.750611-07:00",
"source": "C",
"case_name_short": "",
"case_name": "Thomson Reuters v. Ross Intelligence Inc",
"case_name_full": "",
"judges": "",
"sha1": "a5f8d254d8959f0569d5d8d58583c0549427dc33",
"download_url": "http://www2.ca3.uscourts.gov/Oralargument/audio/25-2153ThomsonReutersetalv.RossIntelligenceInc.mp3",
"local_path_mp3": "mp3/2026/06/11/thomson_reuters_v._ross_intelligence_inc_cl.mp3",
"local_path_original_file": "mp3/2026/06/11/thomson_reuters_v._ross_intelligence_inc.mp3",
"filepath_ia": "",
"ia_upload_failure_count": null,
"duration": 3154,
"processing_complete": true,
"date_blocked": null,
"blocked": false,
"stt_status": 1,
"stt_source": 1,
"stt_transcript": "Thomson Reuters v. Ross They're still setting up over there. Might need some more chairs. Mr. Davies. Two minutes for rebuttal, Your Honor. Sure. Thank you. May it please the court. Under this court's ASTM versus UpCodes decision, unanimously issued on April 7, 2026, the court should summarily reverse this appeal based on fair use. Fair use ruling here brings into question the core technology of the AI revolution. In 2015, a startup called Ross Intelligence built a legal search platform. To train its model, Ross first adapted a headnote into a question. Talk to me about factor four and the impact on the market. So factor four, there's a straightforward answer. You have to look at the statute which says, the effect of the use upon the potential market or value of the copyrighted work. The copyrighted work here is the headnote. There is no market for a headnote. And there's no market for a headnote. Sorry, Your Honor. What do you make of the argument about the market for AI training? The market for AI training, Your Honor. The reason why there's no market for AI training is we are the only ones who used headnotes for AI. And we didn't just take a headnote as raw data. We built them into these memos, which have the two parts. We have the memo itself with the questions, adapted from the headnote. But the key to that memo are the six different answers. Some right, some OK, and some wrong. Because that's how you train AI today. You need that wrong answer. And then from that memo, we train. And the point I'm getting at, Your Honor, is it would be circular if we can come up with such an idea, implement it, and then be accused under factor four of dividing every fair use. Somebody could say, oh, I would have done that. But I guess maybe it's a question for your adversary. But there is some evidence in the record that they were trying to do things like this. The only evidence about AI on the Westlaw side, Your Honor, points to a 2011 algorithm that hadn't been updated since. And it's important for us to get across what AI means. There's AI generally that's been around for a long time, artificial intelligence, anything a human does, tic-tac-toe. What we did is the fundamental AI that is changing the world today. We use deep learning neural networks. So as we point out in our brief, our founder was from the same lab that generated he won a Nobel Prize. So it's not AI doesn't have like this. It's a very broad definition of the market, Your Honor. If you just say, well, it's AI, they were going to use it that way. You have to be much more precise about specific use. Wait a minute. Doesn't West have its own AI-powered general language search engine? You mean today, Your Honor? Yes. I mean, they have to create it, right? So yes, they have it today. Is that correct? I believe so, Your Honor, that today they just spent a lot of money to buy somebody else who is doing it to reach an agreement with a company that can do it. Yes, and that's one of our key points here. That's the same fundamental technology that we invented and that is part of why it's fair, because that's what it's about. It's not a market protection system, Your Honor. Didn't they use their headnotes to train the West Search Plus feature? Their headnotes were used in that context, Your Honor, to just reach other headnotes. They were not part of how that. I don't understand what you're saying. Just to reach other headnotes. It's not a method of finding opinions. But Your Honor, even if they did, even if they did, that's the copyright is not about protecting markets. We have to start with factor one, Your Honor, with respect. The factor one here is the specific use. What did we do with their copyrighted work? And because what we did was so transformational, that narrows the question in factor four. Can you flesh that? You're going to flesh that out. Why was it transformative as opposed to just a different type of legal search engine? To start, we wrote those memos, which is not what the headnote is. We had the memo. We used the headnotes as part of the memo. But the core of the memo was answering the questions. But yeah, let me unpack that, Your Honor. And my question, I guess, is more about the ultimate product. Well, the purpose, the factor one is about the purpose and character of the use. OK, so I'm asking about the ultimate product. The ultimate product has to be the copyrighted work. Factor four doesn't say ultimate product. So you just don't want to answer my question? No, I'm happy to answer your question, Your Honor. OK, so talk to me about the comparison between Westlaw and the ultimate product. The ultimate product is, in some sense, we would be competing with them. I think that's what you're getting at. Yes, we are competing with them. That's one thing that I'm getting at. But I'd like to understand what the differences were. The difference, we, our product, internally operates very differently than the Westlaw product. But it was a direct competitor, correct? I think that's what you're getting at. It's a direct competitor. And that's great. We won competition in this country. The Copyright Act is not an anti-competitive act. It's about looking at the specific use. Your Honor, is it? It aimed to serve as a commercial substitute for Westlaw. Is that not what its internal documents say? Well, external documents, sorry. Yeah, we were competing, yes. We wanted to bring to the country a better legal search technology that takes advantage of today's AI. I'm trying to give you an opportunity to explain on the record what was so different or transformative or earth-shattering about what Ross planned to do as compared to what happens when I log into Westlaw in the morning. At some high level, Your Honor, as in Andy Warhol, there were pictures that looked similar. You could see similarities in the picture. But what we have to do is look at the statute. And the statute has four. There's a lot of dancing going on right now. And I'm just asking some very fact-bound, specific questions about your client's product. Because I am sincerely trying to understand what the product is. And I'm under the impression that for whatever reason, strategic, client-based, I don't know, that you don't want to answer them. No, no, no. So I'll give you another chance. Could you please describe to me what Ross's proposed engine would have looked like and the differences and why it was so transformative from what happens when I log into Westlaw in the morning? From a user perspective, if you're asking to find, you want to find a case, at some level, they're the same. But that doesn't make it a violation of unfair use. So I'm not trying to resist you. I'm embracing your answer. I'm asking you a factual question. And truthfully, I'm not trying to go anyplace with it. I want to understand what your client was trying to do. Our product is you would ask a question in natural language. And then you would get six answers. You would get a stream of answers. But the first, they're all quotes. Here's another difference, Your Honor. Our answers are quotes from judicial opinions. So there's no risk of hallucinations. There's no making things up. The way our search engine works, using the numerical representation, that is important to understand. We teach the legal language connected to the judicial opinion. The question has legal language in it. And what we have taught this machine now is how to think like a lawyer. You get the question, and you go find it. So another key difference, the unusual part of this is we were only trained on 0.08% of all the headnotes. But we can answer any sort of legal question because it learned the semantic relationships in the legal context. Under the first, may I turn to the first factor, Your Honor? How is that transformational from the user's perspective? The question, as you put it, Your Honor, is what the user did with the original work. So let's start. Westlaw's headnotes, they're users' indexes to try to understand an opinion. That's the use. Our original work, we had the memos, which I've described. But we also have this algorithm that I was just briefly mentioning, where we use the latest technology to teach the model how to think like a lawyer. We teach it the language of the law. That has to be transformational. Other district courts have called it spectacularly transformational, have called it highly transformational. And certainly under UP codes, that is just such a dramatic difference. We took something. And another key point, Your Honor, is we don't reproduce the work. A lot of these cases, like the Warhol case, the work is reproduced. There's no headnotes in what we teach the public. That result does not have headnotes in it. It's just quotes from the judicial opinion. And so that is why it's spectacularly transformational. And Your Honor, I think it's also, we took so little. We just took 0.08% is all we copied of 28 million headnotes. And let's also come back to factor two, the nature of this work. As Your Honor's opinion made clear, this is legal context. It's not something that is creative. It's not poetry. It's not like the creative works that are going on in the other district court cases. Well, the bar for originality doesn't seem very high. If we focus not on the ones that directly quote, put aside for a moment, because that's not what the summary judgment was about. If we focus on the ones that are not direct quotes, that take a full paragraph of language and dwindle it to whatever it was, 250 words or whatever it was, or less, why is that not sufficiently original? So first, each opinion has multiple headnotes. I'm aware. OK. Well, if you look at page eight of our reply brief, we quoted both the opinion and what Westlaw did to it. And it's just so little changes. And under the bank's decision. In the one example in the opinion, we're talking about 2,000, I can't remember, it's 248, but it's something close to that, Westlaw notes. I looked at a lot of them. There are some where you have significant language that's changed. And let's be clear about where that number comes from. That number comes from your expert that says, nope, these changed. This is the change. And your argument, as I understand it, is that none of those, I'm not focused on the ones that are quote, none of those are sufficiently original. And I'm trying to understand why. So three responses, Your Honor. First, a judicial opinion is part of the public. What you all write belongs in the public. And the trivial differences that you're discussing, Your Honor, should not be enough to take that away from the public. How are they taking the opinion away from the public? Couldn't you have taken the exact same opinion and formulated your questions? We could have taken the exact and formulated. Yes, but the question is, under the originality, are they doing anything original? Did they write those headnotes? They did not write those headnotes. They copied, Your Honor, the work. So I'm trying to understand how. If you take an entire 10 pages that I write and summarize it in a page, you summarize it in a page, is that copying? No, Your Honor. We're not talking about summaries. We're talking about headnotes. I understand a headnote. So the Seymour headnote, this is one of 10 in a five-page opinion. And there's also the Feist decision that you're on. The Seymour headnote, the part that you're quoting from the brief, the table here, you excerpted in the right column. You use ellipses because you're actually lining up excerpts from that paragraph in Seymour with the headnote. And so when you go back and look at the opinion, there's citations in there. There's things that Wes's editor, to create the footnote, cleaned up, thought about what was salient, and included in the headnote. And this is one that I think you all are talking about as a straight copy. But even in that, there is discretion. So there may be discretion. The only word in Your Honor's question I would comment on is the word creative. This was the work, the instructions that Wes Law gives to its editors is to take out any creativity. Creativity, the reason are there are so many verbatim. It just seems to reflect a judgment about what they wanted in the headnote, and what they thought was salient to communicate in the headnote part of it. Yes, and same true in Feist. It's not like anybody is saying there was an effort and work done in Feist. But the idea here is to get the line. I'm not talking about sweat of the brow. I'm talking about a, I don't know, this is, I think everybody agrees, this is not some remarkable level of creativity. But there are things going on here that go beyond copying, just in the excerpting. But the goal is to track the language of the opinion. If there's some difficulty, you do the best you can. But the goal of the system, and that's what's so lacking creativity here, the goal is to copy the judicial opinion as close as possible. And of course, what we're talking about too. Not what this example shows. This example shows judgments made about what to call out for purposes of the headnote. It's not to copy as close as possible. There's two ellipses. And this is the example you gave to us. This isn't, I think, one of the ones that everybody agrees was discretionary and different. This is a copy. And even in this, there's two sets of ellipses. There's three sentences from the opinion that were called out from a paragraph and condensed into a headnote as part of their thinking and their judgment. And rather than going to Seymour and your client doing it themselves, you took the headnote. So this is their best one. We got this because this is the one they highlighted. But your honor, factor two under the fair use. Even if you think there's some tiny bit of creativity here, which I don't think there is given the goal of the system, it certainly is a factor two of fair use has to be. We have to win on factor two, as the district court agreed. We win on factor three, as the district court also agreed, because it's so trivial. So what we have here is a use where we're ahead on factor two. We're ahead on factor three. If they are ahead on factor one and four, what's the result? Well, Google turns. I believe Google starts with factor two. So it's certainly possible to win. But I don't agree that we should lose on factor two. I'm not asking to agree. I'm asking, if we think they're ahead on one and four, what's the result? We prevail on factor two and factor three. I think, who wins? Who wins? We win, your honor. If you take two and three, and they take one and four, you win? It's a multi-factor test. And another key part here is that copyright law. Is there any factor that's more important, so to speak, than the other factors? It's a equitable decision. You can find language in cases pointing one way and pointing the other way. But the key is, copyright statute came after the common law development here. These are case-by-case determinations. When you have something as spectacular as this AI system, when you have something with a market where nobody is buying headnotes, when you have something with trivial Just because you mentioned equity, in the opposition brief, there's a discussion of bad faith and evidence of bad faith by your client. It's not, I don't think, completely addressed in the reply. You guys have space constraints. And you have to think about what you want to present to us. But could you address it now, whether there are factual disputes about the bad faith evidence and how that fits into the fair use analysis? Your Honor, a couple of things. We did not, we paid for the original database. We paid for the memos. We haven't used any pirating issues. And so I think, first off, we resist the characterization. But let's say the characterization is fair. But what about the communications with Dentons? And I think there's another party about trying to get these in the first instance through them. What are we to make of that? The short answer is that that is way, that doesn't compare to what was done in Oracle versus Google, for example, where people high up at Google were saying we need to negotiate a license. And it doesn't play in. Judge LaValle says copyright is not a privilege reserved for the well-behaved. But regardless, these are issues that can be worked out in the pending litigation of the license violations. They're not part of the fair use inquiry. So I would say nothing, Your Honor. Well, explain to me why that's the case, what you're relying on to say that we shouldn't look at bad faith. So Campbell, being denied permission to use a work is not way against a finding of fair use. The Sega case, copyright is not a privilege reserved for the well-behaved. That's LaValle. I mean, that's just not the way the doctrine has developed. You think those cases stand for the proposition that bad faith is irrelevant, or that the bad faith in those cases didn't outweigh a fair use finding? So in the Supreme Court and Google Oracle, there is a paragraph. And they strongly suggest it doesn't matter. But they don't need to hold it. And I think that would make sense here. Because the goal of this is a copyright case. It's an interesting case. It raises lots of issues. But it's a copyright case. And the point of copyright is progress. And here we have a startup that 10 years ago was engaged in the very technology that is changing the world. Not in the AI, just generally. In the deep neural network technology that was driving change today. It's the same basic stuff. Yes, today there's much more money in it, and much more data centers are everywhere. But it's the same fundamental technology coming out of the same lab. And that's why it would be such a shit. That's why copyright is progress. This is what my client was doing. And it should be recognized. Thank you. Thank you. Good morning. Good morning. I'm J.L. Sindaly. I'm proud to represent Thomson Reuters and West Publishing here. I'm going to hit you with a question right out of the gate. And it's about, what are we talking about? As I understand it, the district court issued a summer judgment as relates to the 2000, you know the exact rest of the number, headnotes. But there is a discussion about the remaining headnotes that may be described as direct copies. Your brief addresses that additional discussion. Are we focused on that? Should we ignore that and just focus solely on what summer judgment was granted on? We don't think you need to address anything other than what we moved on. What we moved on on summary judgment was the 2,834 headnotes, which after, in a very thick document, Judge Bibas carefully looked through and said 2,430. But didn't he certify the question to us as to all headnotes? He did. Two things. Under Calhoun, this court doesn't have to abide by the certification. It can decide the appeal on any grounds that it thinks proper. Makes sense? Pardon me? Would it make sense? I think it would make sense here for you to, we think that what the district court said was right. But the court was talking about, when it talked about all headnotes, was talking about that we have a presumption of validity because we have co-copyright registrations for many years granted by the Copyright Office. And they had to challenge them. And the court in the beginning part of his opinion just said they didn't meet their burden of challenging the presumption of validity. And this makes sense because the Supreme Court held 100 years ago in Calhoun and more recently in Georgia that headnotes are copyrightable. So this court doesn't need to revisit whether headnotes are copyrightable. They are. The law draws a distinction between things that are the work of people writing about law and the law itself. What I thought, so our point is, as Your Honor had pointed out, the headnotes, their own experts showed that the headnotes do not match the judicial opinions that we moved on and do match the bulk memos. And as a result, the court found these were protectable and were infringed. And that, I think, is the proper scope of this appeal. If you wanted to go beyond that, you certainly could go beyond that. But from a categorical point of view, headnotes have already been decided as copyrightable. And the only things that we were all briefing below were these specific numbers. What I'd like to do is start where this court started on fair use and try to clear up some of the both factual and legal questions that came up on that. Because there's a core principle here. In Warhol, the Supreme Court said, substitution is copyright's bet noire. This is a classic case of substitution. And let's start with factor four, if that's OK. So factor four looked at market harm. Harper says it's the single most important factors. So some factors are more important than others. And usually, it's factors four. And some people talk about factor one. I've been doing this a long time. I don't know of a case where someone, in answer to your question, won on two and three and lost on one and four and was deemed to win fair use. I'm sure there's enough clerks in the room that would further answer that question. But I think that's not how the law works, given how the court has said that factors one and four are most important. On this factor, I think the parties are sort of worlds apart on what the relevant market is and what we should be looking at. Can you flesh out your side and the authorities for it? Sure. Well, first up, it's their burden. Let's remember that. This court made clear in upcodes that it's their burden to establish a lack of market harm. And the Redigi case is helpful because it says, when a secondary use competes in the rights holder's market as an effective substitute for the original, it impedes the purpose of copyrights in incentivizing new creative works by enabling their creators to profit from them. And the key point here is not only is their work a substitute, but they admitted it and marketed it as a substitute. The ad that we put in our brief, choose Ross or Westlaw, you can't get a more quintessential example of I'm putting out a product that's substituted. In addition- The competition piece seems a little disconnected from the use that we're talking about. I think that's your adversary's point. And that's why I've been asking about the market for AI training and where that fits into this. Well, there's two kinds of, well, there's three kinds of harm that we've talked. I'm not sure if I'm answering your question, Your Honor, but they're marketing the product as a substitute. And there's lots of documents, some of them I can't all talk about here, where they're trying to take market share and there was actual substitution and all sorts of things like that. That alone is enough for us to win. There's also harm to our market for exclusivity of our training. My friend on the other side seemed to give the impression that we didn't train on our own AI, but in our own content. But as we said, I think in page 12 of our brief, we actually trained on our head notes four years before they even were founded. And we do have an AI tool and we're allowed to use our own exclusivity to build our product. And then the third thing is there's also a potential market, which is also allowed to be recognized under the law for AI training. And we know this, one, because we've used it ourselves. There are other legal research AI companies that would find it useful. And Ross's own conduct here shows that there's such a market because they paid money when they couldn't get it to us, they paid money to someone else in order to get the material. So is it your position that we should be looking at all three of those markets in thinking about whether Ross has carried its burden with respect to each? I mean, it seems like there's a distinct legal question about what market factor four asks us to think about. And I'm not sure that I agree that there's a burden there as opposed to just who's right. And then I do agree that there's an evidentiary burden once we fix the parameters about who wins on the factor. Well, I think we can all agree that fair use isn't a primitive defense and they have the burden, again, up codes of establishing that. And our point is, even though the burden wasn't on us, we actually articulated three types of market harm and they haven't come forward with any evidence to dispute any of it. Almost they said was that, oh, the market really isn't Westlaw, it's for headnotes. But that makes no sense because what headnotes are part of Westlaw, that's how you get licensed to headnotes. That brings another case to mind for me, the video pipeline case, the one involving Disney. As I understand it, there was no market for Disney trailers standing alone either, but the website that stole the trailers was found to have caused diminished value to Disney's website because Disney used that to attract people to their website to do other things on their website. And I'm wondering if you could talk a little bit about that case and whether or not it applies here. It does apply here. It's one of the leading third circuit fair use cases. And again, the court said, similar to Warhol, doesn't even have to be perfect substitution, but I will get at the fact that they're selling the same product, does the same thing, and there's nothing transformative about it. But video pipeline, again, they came up with their own kinds of trailers. Disney had its and others had their own kinds of trailers. And they said that you're gonna hurt my market for what I'm doing with my trailers. And also the idea that once people engage with them, as you said, they will otherwise be interested in other things we're marketing or selling on our website and the like. And once you divorce, they take our content, our question and answers pairs. Remember, they had the cases, their own experts admitted that they could have done their own analysis themselves and come up with question and answer pairs to do the same thing. They didn't need to copy us. They just wanted to. In that situation, there's no justification, which is a lot of what Warhol says. I hope I answered your question, Your Honor. But so the point is on market harm is that it interferes with markets that we went out of our way to identify even though it's an affirmative defense. But let's go to factor one and your questions about transformativeness and what's transformative about it. Because a couple of things, they admitted in the undisputed joint statement of facts, docket 796 at 01, that Ross's legal research platform is commercial and for profit and that they're competitors. So the whole idea of commerciality, which is very important in looking at factor one, they've admitted that they are commercial. In addition, on terms of transformation, in Warhol, the Supreme Court said you need to consider the purpose of the use and where the purposes are the same or similar, the use is not transformative because the use of an original work to achieve a purpose that is the same as or highly similar to that of the original work is more likely to substitute for or supplant the work. And the district court held that Ross's use was not transformative because it doesn't have a further purpose or different character. As the court put it, it was using the head notes to create a legal research tool to compete with Westlaw. I feel like there's one version of this that my guess would be that you like where it's the head notes are sort of a cheat sheet that your client designed to opinions that everybody had access to and Ross could have just gone to the opinions, they took a shortcut and you guys have your rights. There's another version of this that you hear from your friend on the other side that they took the head notes to train up a transformative technology that is just materially different from what you guys do in terms of, just for simple me, there's a search bar and what I type into Westlaw versus what I would have typed in to Ross. What is in the record about sort of the precise nature of that training and the technology that we should be thinking about? Well, the training, it's just a means to an end. It's just a means to copying. What you look at is what the purpose is under Warhol and this here, the purposes were the same to come up with dueling legal research platforms. Now, you asked my friend on the other side, well, how different was what your product did compared to what Thomson Reuters plotted? With him, I was more focused on the end use and with you, I'm more focused on a comparison between what Westlaw is doing with the headnotes in its search functions and what's in the record about this and I'm not asking for any state secrets, but compared to what the Ross product would have done after the training. Well, what's in the record, I believe it pages 12 and 13 of our brief is that we trained on our own headnotes to create AI products that we offer to the public. You said that started four years before Ross was even created? Correct, correct. And so as much as they try to cast themselves as tremendous innovators, they really were late comers in terms of what Ross and it's not just us, if you look at the amicus brief of Lexis, they also had an AI product in 2010. So that's what the world was, but the question is, what did this product do? And as the district court said, what it did is it spat back cases, that's what you heard, it wasn't like generative AI, I'm gonna write a brief for you or anything like that. It was like spitting back list of cases and quotes, which is at least when I use Westlaw, that's part of what you get from Westlaw too. And I think it's very telling that in Bart's philanthropic, while the court there was very comfortable finding that what was transformative was the AI generative issue in front of it, but it agreed with the district court here that as it put it, using a proprietary system for finding court opinions in response to a given legal topic to train a competing AI tool for finding court opinions in a response to a given legal topic was not transformative. In other words, you can't just say AI and have that just be across all fact patterns. The whole point of fair uses the Supreme Court has recognized is it's fact specific. You look at the uses and you look at whether in fact there's substitution and justification. And here, there's a direct. What's your best evidence on the record of market harm to your client? Well, the best evidence on market harm is one that's their burden to show that there isn't, but our evidence is that they advertise this as a competing product, as a substitute for us, and that they admitted that they in fact did succeed in taking market share for us and causing people to switch. The other key thing that my friends on the other side don't even mention in their opening brief and our view gives short shift to in their reply brief is the concept of widespread use, because the Supreme Court is clear that you don't just look to what Ross did from a fair use point of view. It's not a damages analysis. It's an analysis of everybody and his brother who were allowed to do this. What would be the consequences? They don't mention that, and with good reason. They don't mention it because the net effect of it is that you end up with, they don't mention it because if everyone was able to do this as both our brief and the Lexis Amicus brief said, why would we continue to be hiring and training people to do this and analyze the law and be useful for people if you can free ride off of that? I see I have a red light. I just wanted to, if I could, just say two quick things on factors two and three. Factor two is that this was the creative nature of the headnotes. We all know that if we all analyzed a case and were told to headnote it, we would probably do it differently. That, given the nature of the use here, that this was for a legal research platform, these are creative uses, and they admitted that they wanted those pairs. With regard to factor three. On factor two, I think you talk, and I think you just referenced it, the investment of time and funding in the editors, and you cite, there's a Ninth Circuit case and a Second Circuit case, and you guys battle about whether those are opposite. We haven't looked to this consideration in this circuit. Is there anything else you're hanging your hat on for we should be looking at that? The one thing I can add on, thank you for asking the question, Your Honor, is that recently, it's a District of Delaware case, but in Deloitte Consulting versus Sagitek Solution, Judge Bryson of the Federal Circuit, sitting by designation in the District of Delaware this year, adopted Waldata's reasoning that while hard work doesn't go to anything about copyrightability, it is something to consider with regard to assessing the factor two. And then, if I can turn to factor three? Okay. And then in terms of factor three, the key thing here is they took the heart of our work. You heard a lot of comment of, oh, it was only .08% or something like that, but the heart of the work, what Westlaw is, what differentiates it from merely a collection of cases that you can get, which they did from other vendors, is the editorial content that you can use the West Key Numbers system, you can use the headnotes, you have different choices as to how you wanna research, but ultimately, it solves the problem which their own expert admitted was a really good thing of solving the problem of finding the needle in the haystack. That's one thing. The only other point, and they took a lot, it was in bulk memos, each one of them had a headnote, and that's not like one or two lines like in some of the Salinger cases. This was a lot of taking. But the other thing I wanted to just point out that the one thing that I don't think was quite right in the district court's opinion on factor three is the court said, oh yeah, it does look like they took the heart of the work, but gee, in the Google Books case, the court focused there on factor three on the fact that things weren't visible. And since the headnotes weren't visible, the court said, well, maybe that should go towards Ross. But the reason that doesn't apply here is the Google Books was all about the idea of whether Google Books, the book search tool, substituted for actual reading of books. So what the court said in its factor three discussion was no one's gonna read these little snippets as a substitute for reading whole books. Rather, Google Books is a search tool that lets you find books, and that's a good thing. That concept of visibility is not relevant here where the whole point is the district court found, and they don't dispute that the whole engine, excuse me, the whole Ross platform was built on Westlaw. And that's a difference. And that's my last point. I've got one more. I want you to talk a little bit about the cases involving intermediate uses, so Sony and Sega and the Ninth Circuit, and then the Supreme Court's Google decision in the relation to declaring code, like compare headnotes to declaring code. Okay, let me start with Sony and Sega, and I will try to remember. I know you want me to do Google next. I'll remind you, yeah. So if I forget, I have a feeling you'll remind me somehow. But all right, so Sony and Sega. Those cases actually go our way. Why? Because they cite those cases as if they stand for the proposition that intermediate copying is fine and that you can do it and that's not a problem. Actually, they say the opposite. As the Sega court said, the Copyright Act doesn't distinguish between unauthorized copies of a copyrighted work on the basis of what stage of the alleged infringer's work the unauthorized copies represent. And so the Copyright Act unambiguously encompasses and prescribes intermediate copying. And we know this in the Copyright Amicus Brief of Copyright Professors talks about this because Section 106 has separate rights for distribution and reproduction. They're basically saying unless there's a distribution, it doesn't matter. But that's not what the Copyright Act has written. And thus, the Sega court said intermediate copying of computer code may infringe exclusive rights regardless of whether the end product of the copying also infringes those rights. So that's what the court said. Then the court went on to say in those cases that because you're dealing with computer code, we're the only way you could understand the functional nature of that code to create a compatible product. In one case, it was cartridges that could play on the Sega station. In the other case, to be able to play the games on a PC. Because of that compatibility need, you needed to be able to do that intermediate copying to achieve that purpose. That is completely different from this case. There's no argument whatsoever of any compatibility need, which also relates to why this is an Oracle v. Google. And it also just highlights, again, that there's no need. The whole point of the Sony and Sega case is you had to reverse engineer in order to understand the functional aspects. As we've heard and my friend admits, they didn't need to copy to do what they did. So those cases do not help them. If anything, they help us. And with regard to Google v. Oracle, that case- And the declaring code. Oh, excuse me. So Google v. Oracle, in contrast to, go back to factor two a minute, in contrast to the situation here, which is a literary work written by humans where someone is analyzing the law and deciding what headnotes to write and what case passages to link them to and the like, the court in the second factor was very focused on APIs declaring code as being basically the most functional kind of code as it could get. So in that situation, it was the most extreme situation you can think of where factor two existed where there was essentially, just maybe a little bit over, the court found copyrightability there, of course, but not extensive work. That's not the situation here where there's no computer code, don't, no, I can tell you're asking me a question. There's no ones and zeros, but at least I think in the way end users of Westlaw consume opinions, there is a functionality part of this that there's sort of, the argument would be there's a row copying of a headnote that then in the same way that a declaring code sort of shoots the programmer to the right part of what they actually wanna get at and the more creative aspect of it, the headnote, you click on it, it shoots the user to the part of the opinion that is public and is everybody's. I disagree with you, your honor, because the whole point of, after the court said declaring code was this special kind of lowest level of computer code, it really based the thrust of its opinion was on this compatibility desire to let programmers who are already familiar with those APIs that were just barely protectable according to the court to be able to use them in an entirely new platform. A platform, by the way, where the court said under factor four, there wasn't market harm because Oracle had not entered or didn't seem likely to enter the market for using Java on mobile devices. Totally opposite, this situation. And I think your other question, did I answer your other question? You hit my three cases. I hit your three cases. I'm good, yeah, thank you. Three cases. My bottom line, your honors, is, and thank you for giving me a couple extra minutes, my bottom line, your honors, is I go back to the Constitution. The Constitution is designed to give people copyright protection so that they could innovate and come up with, invest and come up with works. And that's what my client has done for over 100 years. It was once a startup too. And we all know, some of us went to law school when there was paper that we used to shepherdize cases. And then Westlaw came around and suddenly that was different. And now we're using AI and we've been using natural language search for before them, as has Lexis. That is all possible because of copyright. And the idea that someone could come along and can't, oh, it's AI, is just wrong fundamentally under the law, under common sense. And yes, I'm happy to talk about bad faith if you wanted to talk about it too. But if ever there was a case where that mattered, it would be this case and it favors us. Thank you. Thank you, your honor. One place where I agree with my colleague is you cannot just say AI. We're gonna give some sites because if there's one thing I can leave the court with, it's to convey the complexity and sophistication of the technology here. So we lay it down in our brief at pages 10 to 14 at appendix 6992 to 7000. That's where the fundamental version of AI that we invented, that we use, that is now changing the world, the chat GPT is the anthropic, that's the technology we're talking about. We're not talking about the AI from 2011 that was never updated. Now yours is not generative AI, right? It's not generative AI, your honor, but that word, as one of your amicus briefs points out, has no technical meaning. We don't output, but we could have output. But you don't, you don't spit out, I can go on and ask chat GPT, write a brief about this. May hallucinate some stuff, but it'll write it. I can go on and say, can I legally do this? Might not be completely right, but it'll give me an answer. I can say, you know, I've been sued for X, Y, and Z, what can I say to convince my insurance company to cover me? And it'll give me an answer. That's not what your product did. No, no, and that I think reflects on the integrity and knowledge of the founders, because they didn't do that. They could have had it generate answers, but they went to the judicial opinions, because one of the visions of this company is to get people access to law, not to generate hallucinations. But you ask a question and it gives you quotes from a case. Yes, yes, Your Honor, and that the sophistication is it links your question to judicial language, even though we only use 0.08%. So what the training lets you do is answer a legal question that the model has never seen before. And that's why we didn't take the heart. We could have used any 0.08%. You know, any 0.08% we could have used. Your Honor asked about summary judgment motions, A58 is where the district court below denies our summary judgment motion on all the head notes. And Your Honor, you've asked a lot about the market. I would refer you to both the Oracle case and the Andy Walhart case, where it's not the market in general. It's a distinct market. In Oracle at 1207, Your Honor, Google's Android platform was part of a distinct and more advanced market than the Java software. And so we don't collapse the markets. The AI that my colleague was talking about that they had just went from head note to head note. It didn't go from head note to opinion. And that's at A8615. The record here of the market, Your Honor, completely lacking. Even if we had the burden, they do need to come forward with some evidence that the market has been harmed at a significant level. And they've shown- It's just precedentially established that it's your burden. Yes, sure, it's out, yes. When you, on rebuttal, say they have to come forward with some evidence, what is the basis for that? Because we have argued, we've made an argument. They need to refute the argument. It's the record. And UpCode itself doesn't turn on burden. It turns on the fact that the evidence there had no- I think the way to frame it would be, has your client established that there is no market harm? Right, but as a practical matter, it's Westlaw's subscriber numbers. It's Westlaw's customers. How on earth can we do that? I mean, we did our part, but they haven't shown any evidence. In fact, there's actually evidence the record's shown the other way, that people left Ross to go back to Westlaw. They have an obligation under UpCode to give concrete numbers, not sort of anecdotes and ads that don't even mention headnotes. They have to come back with real numbers. You know, I mentioned the video case. That was a reproduction case. Another point here is we don't reproduce the headnotes. It's not as if the customers in the market can get Westlaw headnotes through us. We just give them the opinions as we were talking about. And then on the Sega Sony cases, those actually state the public policy that I think is so fundamental here. Fair use is not an invention. If I'm an author, I may not love fair use, but it's what Congress decided. We have to allow people to do things that make sense. There's no strict necessity requirement, and it's fine if you're using it. What you have to have is a legitimate purpose, and we couldn't ask for a more legitimate purpose. I think Sega uses language, humans cannot read the object code. Humans cannot read the object, yes. And Ross employees could read the opinions. Ross, the opinions are not the work, though. Right, but my point is it was not necessary to scrape the headnotes to get to the legal opinion. Right, and there's no necessity requirement. Well, you're relying on these intermediate use cases, and those are situations where it was without the intermediate use that we're talking about, they couldn't access the information. So, Your Honor, the Oracle Google case, that there's 37 APIs that Oracle wrote, Google was copying them. Right, so that's a different line of authorities, but in the Ninth Circuit cases that we're talking about, we got ones and zeros that had to be reverse engineered in order to create the transformative uses that those courts talked about, which seems to be fundamentally different from what you're telling us the product is, is a lawyer puts in a question and gets an answer linked to case law that everybody agrees is available, is fair game. And you guys took an intermediate step, and when you cite the Ninth Circuit cases, Sega and Sony, I think the argument is we should be looking at those steps that they took to modify those platforms as similar to what you did, and I just don't think it is. I mean, those cases are about public policies, and they say that, and here's a quote from Sega, when the person seeking the understanding has a legitimate reason for doing so. We were under huge time pressure, these startups don't have a lot of time, they have to raise money, they have to move fast. That sounds very commercial, when you're talking about public policy. Your Honor, startups definitely have a commercial element, for sure, but in America, certainly in the AI context, we have what we have, because of the energy that's coming out of that community. Yeah, I think this case is about balancing what we have, and where are the right lines. It's all about balancing, as all copyright cases are, Your Honor. Yeah, if there are no further questions. Thank you. Thank you. Counsel, before we leave, I'm gonna ask you to check in with the clerk, because we'd like a copy of the transcript, and I'll ask you to split the costs, and the clerk can explain to you how you go about doing that."
}