Latest / CLEs You Actually Want to Hear / Prepublication Review in the Age of AI
Transcript
- 0:03Hello, welcome to a pre -publication review in
- 0:07the age of AI. I'm David Shraga, Secretary of
- 0:09the Beverly Hills Bar Association's entertainment
- 0:12law section. I'm very pleased to be with you
- 0:14here today. Before we get started, we wanted
- 0:17to acknowledge Lawyers Mutual Insurance Company,
- 0:19a long -standing sponsor of the entertainment
- 0:21law section. For five decades, Lawyers Mutual
- 0:24has been the only California -based company to
- 0:27continuously provide lawyers throughout the state
- 0:29with professional liability insurance coverage.
- 0:31Their experience and expertise are complemented
- 0:33by the fact that LMIC is directed by a group
- 0:36of dedicated lawyers themselves. That means they
- 0:38are uniquely skilled and have folks on the team
- 0:40who are among the very limited number of certified
- 0:43legal specialists in the state of California
- 0:45in the area of legal malpractice. And as an added
- 0:47benefit, that specialized team operates a hotline
- 0:50that's available to policyholders. who have pressing
- 0:52questions that they need immediate help with.
- 0:54They offer special rates with competitive premiums
- 0:57for a wide range of areas, including entertainment
- 0:59law. We have come to know these folks here at
- 1:01Beverly Hills Bar very well. They are a great
- 1:03partner, and you can learn more about them at
- 1:05LoadersMutual .com. I'm now delighted to introduce
- 1:08our highly accomplished panel, Dale Cohen, Chris
- 1:11Perez, and Amanda Harris. Professor Dale Cohen
- 1:14is the Director of the Documentary Film Legal
- 1:17Clinic at the UCLA School of Law, where he leads
- 1:19student clinicians in providing pro -bono legal
- 1:22services to documentary filmmakers. And he serves
- 1:24as Senior Special Counsel to Frontline, the award
- 1:27-winning PBS documentary series. His extensive
- 1:30career as a media lawyer, litigator, and news
- 1:33executive has spanned organizations, including
- 1:35NPR, Radio Free Europe, slash Radio Liberty,
- 1:38Cox Enterprises, and The Tribune Company, as
- 1:41well as litigation partnership at Sonnenschein,
- 1:44Nath, and Rosenthal, which is now Dentons. He
- 1:47earned his JD cum laude from Northwestern University
- 1:49School of Law and is co -author of the textbook,
- 1:51Media and the Law, currently in its second edition.
- 1:55Next up, we have Chris Perez. Chris is a partner
- 1:58at Donaldson Calif Peres, where he represents
- 2:00independent producers, production companies,
- 2:03writers, and other creative clients across all
- 2:05aspects of film and television development, finance
- 2:08and production, clearance, and distribution.
- 2:10He is co -author of the fifth edition of Clearance
- 2:12and Copyright, the industry's leading text on
- 2:14right clearances. And he's been recognized by
- 2:16Variety, The Daily Journal, and California Lawyer
- 2:19for his work on landmark DMCA exemptions for
- 2:22filmmakers. Chris also serves as a board member
- 2:24of the International Documentary Association
- 2:26and as an adjunct professor of law at USC Gould
- 2:31School of Law. And finally, we have Amanda Harris.
- 2:34Amanda is an associate attorney at Jassy Vic
- 2:36Carroll and LLP. Prior to joining the firm, she
- 2:38served as a business and legal affairs exec at
- 2:41a premier talent agency where she provided global
- 2:43representation to clients across film, television,
- 2:46publishing, digital media, commercial endorsements,
- 2:48audio, theater, and book to screen adaptations.
- 2:50Before that, she was an associate at a widely
- 2:53respected private law firm here in Los Angeles
- 2:55where she maintained an active pro bono practice.
- 2:57alongside her commercial litigation work. Amanda
- 3:00earned her JD from Harvard Law School in 2023,
- 3:03where she was recognized with the Dean's Scholars
- 3:05Award. She received her BA with honors from Stanford
- 3:08University in 2019. Amanda is also the co -chair
- 3:11of the BHBA's AI in the Law section and the organizer
- 3:14of this great panel. And as for me, I am privileged
- 3:18to moderate this conversation today and being
- 3:20the company of these highly accomplished folks.
- 3:23For my part, I am a mediator at the Future Resolution
- 3:26with a background as an entertainment litigator
- 3:28in private practice, a business and legal affairs
- 3:30executive in the Hollywood studio system. And
- 3:33I've also enjoyed handling a range of business
- 3:35and employment disputes and settlements along
- 3:37the way. So with that, let's get started. First
- 3:39of all, thank you everyone for being here today.
- 3:41It's really a pleasure to have you. Thank you.
- 3:45Thanks, everyone. Thank you. Thank you. So OK.
- 3:49So let's just get started today by giving our
- 3:53audience kind of a quick map of where we're headed
- 3:56today. And to extend that analogy, for lawyers
- 3:59who are working with filmmakers and vetting documentaries
- 4:02where AI is involved, we really are in uncharted
- 4:04territory. Each of you is at the forefront of
- 4:08this area. in overlapping yet distinct ways,
- 4:12and I think that the combination of your experience
- 4:13really provides a terrific perspective on this
- 4:16topic. Certainly there's no shortage of conversation
- 4:19about AI in Hollywood these days, but we're here
- 4:21to focus on the documentary space. To frame this
- 4:26discussion for folks working in documentaries,
- 4:28we know that for creators and practitioners,
- 4:31there's very tangible and very public implications
- 4:33to how AI is adopted in their work and to how
- 4:36they create and also relate to their audiences.
- 4:39There's also an equally important practical aspect
- 4:41of this that affects how we think about clearance,
- 4:44establishing rights, working with AI, and distributing
- 4:47this work so it can be seen. And that's really
- 4:49where our lawyers come into play, of course.
- 4:53So today we're going to talk about what lawyers
- 4:55and filmmakers need to be aware of and potentially
- 4:57be careful of, how they protect themselves legally,
- 5:01and also how they protect the integrity of this
- 5:03work. They're all really very privileged, I think,
- 5:05to feel that we can support and be a part of.
- 5:07So to start our discussion today, maybe we could
- 5:10get into a real world example from the documentary
- 5:13world that I think will give the audience a very
- 5:15tangible sense of the issues that AI is presenting
- 5:18in our community here today. Amanda, why don't
- 5:21you set the stage by telling us about the what
- 5:24did Jennifer do controversy from 2024? Absolutely.
- 5:29Thanks, David, and thanks to my fellow panelists.
- 5:31This is going to be a great one. So yeah, I think
- 5:34I thought we'd start off to illustrate sort of
- 5:36what's at stake for filmmakers and various stakeholders
- 5:39in these documentary process. The What Jennifer
- 5:42Did example illustrates a lot here. So for those
- 5:46who don't know, What Jennifer Did is a docu -series
- 5:49on Netflix. It highlighted the case against the
- 5:54main person in the documentary, which is Jennifer,
- 5:57who is alleged to have murdered her parents along
- 6:00with three co -conspirators. The real controversy
- 6:05for our purposes is a couple of people online
- 6:08after the documentary came out on Netflix and
- 6:10started receiving a ton of views Accused the
- 6:13documentary makers of manipulating a couple of
- 6:16key images in the docu series with AI People
- 6:20there was an extensive online discord about backgrounds
- 6:24that looked sort of different, or smiles that
- 6:26looked off because one tooth was much more elongated
- 6:29than another. And it really caused a lot of backlash
- 6:32for all the folks involved with the program,
- 6:34for Netflix itself. And I mean, in the interest
- 6:36of fairness, so Netflix, as far as I'm aware,
- 6:38has never commented publicly on this controversy.
- 6:42The executive producer of the docu -series denies
- 6:44any AI manipulation, but just the extensive discussion
- 6:47and discourse around this highlights sort of
- 6:50the stakes for people involved in these projects
- 6:52and raises an issue that I think we as lawyers
- 6:55and counsel should look out for. Yeah, I mean,
- 6:59it's a fascinating example because there were
- 7:02aspects of this, as I understood the controversy,
- 7:05where the filmmakers were trying to reflect something
- 7:08like an emotion, for example, that they would
- 7:10make the subject look happy and sort of in the
- 7:14process of trying to tell the story. But I think
- 7:16really, you know, what it comes back to and comes
- 7:19down to for documentary films is we're really
- 7:22talking about an issue of transparency and trust
- 7:25and how that can implicate sort of the relationship
- 7:30with the audience and also potentially lead to
- 7:32a backlash from the audience, right? So I guess...
- 7:36Chris, maybe we could start with you. You've
- 7:39worked on some very distinguished documentaries,
- 7:41of course, and you're very much in this conversation
- 7:43with clients as they grapple with these issues.
- 7:47As Amanda explained, there wasn't disclosure
- 7:50in this case. There was controversy after the
- 7:52fact. If disclosure is often or not necessarily
- 7:57strictly required, how do you advise clients
- 7:59on when and how to be transparent about how they're
- 8:02using AI? Yeah, it's tough because this changes
- 8:05day to day, right? What was OK to an audience
- 8:08maybe six months ago, or not OK six months ago,
- 8:12maybe they've gotten used to it because of the
- 8:14dramatic increase in the use of AI six months
- 8:17on. But where we're still living right now, I
- 8:20think, is in an age of uncertainty around the
- 8:24use of the technology, but also how it might
- 8:27appear, and particularly in a space like documentaries,
- 8:31right, where people Regardless of what platform
- 8:34it appears on, if it looks like a doc, it smells
- 8:38like a doc, they expect certain standards to
- 8:41be upheld. And, you know, I looked at a lot of
- 8:45the comments on what Jennifer did in terms of
- 8:48the backlash, and it was all over the map in
- 8:51terms of why people were upset. But I think people
- 8:54were mostly upset because they just expect authenticity,
- 9:00right? So as a filmmaker, These are difficult
- 9:04decisions to make, right? Because on the one
- 9:06hand, you could have a situation where maybe
- 9:09the photograph has a blemish, right? And something
- 9:12in the background you don't think the audience
- 9:15needs to see. Okay, AI can do a great job of
- 9:18removing that. But at the other end of the spectrum,
- 9:21you have very obviously wrong things, right?
- 9:24You make the person do something that they didn't
- 9:26actually do, in which case there is potentially
- 9:29a legal claim like defamation or false light.
- 9:32But the middle area is the most interesting because
- 9:37filmmakers now have the discretion to do so many
- 9:39little tweaks that maybe, in their opinion, doesn't
- 9:43really change the basic nature of the truth as
- 9:46they're presenting it. And in those cases, there
- 9:50may not be any kind of legal remedy for the person
- 9:53in the photo, but the audience definitely expects
- 9:58to know that that is happening. And I think that's
- 10:01still the case. The APA, I think we've mentioned
- 10:05before, has a great set of guidelines on how
- 10:08to be transparent. But I don't think that at
- 10:12this point, there's agreement on exactly how
- 10:15to be transparent. Some people say you got to
- 10:18put something in the lower third, almost like
- 10:21this is a reenactment that you've seen in years
- 10:23past. Other people say, well, you can just put
- 10:26it in the end crawl with everything else. But
- 10:29most people would say, well, people don't watch
- 10:31the end call anymore, right? So that's not enough
- 10:34transparency. So I don't think there's one right
- 10:38answer in this regard, but I do think that it
- 10:42is essential that attorneys, number one, know
- 10:47that their clients are actually engaging in the
- 10:49use of AI with particularity, and then having
- 10:52a conversation with the filmmaker about what
- 10:55creatively they're willing to do in terms of
- 10:58transparency. there must be a real tension between
- 11:02the need to craft a compelling story sometimes,
- 11:05right, versus the need to maintain the integrity
- 11:07of the work. And Dale, I saw you nodding your
- 11:10head when we were talking about the critical
- 11:12importance of trust within the community and
- 11:15trust between the filmmakers and the audience.
- 11:17And I'll just add to that that, you know, I think
- 11:20your perspective on this is very interesting
- 11:21because you're working with students at the UCLA
- 11:24film clinic, and I understand that the way the
- 11:26clinic works is filmmakers are paired with teens
- 11:29of law students, follow them through the process
- 11:32and advise them along the way. But we're talking
- 11:33about a generation that's grown up with social
- 11:36media, that's grown up with these AI tools and
- 11:38has a different relationship, I think, to authenticity
- 11:40in the overall marketplace. How does that play
- 11:42out in how you're advising clients and how your
- 11:45students are working with filmmakers when it
- 11:47comes to these issues? It's interesting because
- 11:50for the most part... even the younger generation
- 11:54is looking for the same kinds of transparency
- 11:56and credibility in documentaries that Chris and
- 11:59Amanda and you have already been talking about.
- 12:02So when I first got started with this clinic
- 12:06and I was working with Frontline, which is kind
- 12:08of the gold standard of journalism in documentary,
- 12:14one of the concerns my colleagues at Frontline
- 12:16had was you're going to be working with these
- 12:17newfangled filmmakers and they're not going to
- 12:20have the same kinds of How are you going to do
- 12:22both of these things? And I'm pleased to say
- 12:26for the most part that giving the same kinds
- 12:30of advice to the newfangled filmmakers and the
- 12:33traditional journalism filmmakers in the documentary
- 12:36sphere is well... received by both sets of the
- 12:41filmmakers because they are concerned about what
- 12:44Amanda and Chris have already spoken about, which
- 12:46is credibility in the marketplace. But the industry
- 12:50has also got its own way of enforcing transparency
- 12:54and ethical storytelling. As Chris mentioned
- 12:57in our pre -meeting. Studios, distributors, platforms,
- 13:03insurance companies are all starting to ask questions
- 13:06about the use of AI, and they want specifics.
- 13:09They want to know what's being presented as it
- 13:12really was, what's been created synthetically,
- 13:16and how were those decisions made, and how can
- 13:19they evaluate what the risks are? And so I think
- 13:23all of us would say that we're very fortunate
- 13:26to still be even in this age of AI, that ethical
- 13:30storytelling is still the most important thing,
- 13:34both for the credibility of the film and also
- 13:37for protecting yourself against legal claims,
- 13:40whichever legal claims there may be. And it is
- 13:43true, there's some tension there sometimes with
- 13:45filmmakers who say, gee, I don't want to ruin
- 13:47the shot by having to do this, that, or the other
- 13:50thing. But I think... the history and experience
- 13:54suggests that the best way for them to defend
- 13:57themselves for us as lawyers to help them defend
- 14:00their their films is for them to follow really
- 14:03good ethical storytelling which means transparency
- 14:07with all of their partners as well as the audience
- 14:10but also really really solid fact checking and
- 14:13vetting by lawyers by fact checkers to make sure
- 14:17that that what's being presented it is in fact
- 14:20credible. Yeah, it kind of comes back to this
- 14:25notion that, in some sense, the more things change,
- 14:28the more they stay the same, right? The same
- 14:30fundamental core values that I think have probably
- 14:32animated your conversations with clients over
- 14:34many years are still present there. But as Chris
- 14:39mentioned, the standards are evolving, and they
- 14:42are changing. And Chris, I just think it's so
- 14:45interesting. You mentioned the APA, the Archival
- 14:48Producers Alliance. I know you're a member of
- 14:49the advisory board for their generative AI initiative.
- 14:53And so for folks out there who are looking for
- 14:56guidance about how these standards are evolving,
- 14:59you know what is happening in that conversation
- 15:02you know because the the law seems to always
- 15:05be behind the dialogue right in terms of what
- 15:09we are actually uh the guidance that is out there
- 15:12you know we can't necessarily wait for the courts
- 15:13to rule on all these issues when claims come
- 15:15up um what's emerging in terms of standards and
- 15:19what are some resources yeah yeah i just i think
- 15:21the apa is doing such important work um for exactly
- 15:25the reasons that you mentioned is that Yes, the
- 15:28old adage that the law is always a few years
- 15:30behind the technology. I almost feel like that's
- 15:34just not enough these days. It's like not emphatic
- 15:37enough. It's happening so fast. Their courts,
- 15:41as soon as they get to the point of making a
- 15:43decision, the technology has changed again. And
- 15:47so I think that the work of the APA fundamentally
- 15:50is trying to, first of all, educate people. um
- 15:56specifically film producers and other people
- 15:57in the industry about the importance of transparency
- 16:02right because people are using these tools whether
- 16:05we know that know it or not right i've worked
- 16:08on many films i'm sure where filmmakers were
- 16:12using ai tools and they just didn't tell me right
- 16:14because they were afraid which we can talk a
- 16:16little bit about that but later but i think it's
- 16:20education right um letting people know that these
- 16:25The use of these technologies has real consequence,
- 16:31especially if audiences are unaware that they're
- 16:33being used. But it's also generating an understanding
- 16:37of the importance of real archival materials.
- 16:42When I say real archival materials, I mean, if
- 16:45you're making a film in the civil rights movement
- 16:47and there are particular photographs that depicted
- 16:49certain events or archival footage that you know,
- 16:53that was the only videographer there that day,
- 16:56that material has real value in our culture.
- 16:59And I think the APA has done a great job of broadcasting
- 17:03to the general public that that has to continue
- 17:06to have value, right? At what point is this AI
- 17:10generated material going to substitute for the
- 17:13real thing? And when it does, our audience is
- 17:16aware when it's happening because there's no
- 17:18turning back. People are going to create archival
- 17:20materials using AI. But I think helping viewers
- 17:26understand the distinction between those two
- 17:28things, between that videographer was there versus
- 17:32this was Sora or something like that. Just educating
- 17:38people, I think, is the most important thing
- 17:39that they're getting into. It's a really interesting
- 17:42distinction between. the actual archival material
- 17:45and sort of the recreation of a historical event,
- 17:48because you mentioned previously the role of
- 17:51reenactments and how often that comes up in documentary
- 17:54films. And Amanda, this one's for you. You've
- 17:58got a unique expertise in the sense that you're
- 18:00actually litigating these cases as well. I know
- 18:02we have some folks with litigation background
- 18:04here too, but you're sort of the current trialer
- 18:07in our panel more so, I think. And so I guess
- 18:10from your perspective, How do issues like transparency,
- 18:16disclosure, dealing with reenactments, how does
- 18:19that play out in the context of litigation? Are
- 18:21filmmakers more concerned about that exposure
- 18:24or the authenticity issue, or is it a delicate
- 18:27dance, I imagine? Yeah, it's a great question,
- 18:30very much a delicate dance. I mean, I think Chris
- 18:34and Dil both alluded to this to different degrees,
- 18:37but with this issue, a lot of... A lot of the
- 18:40concern isn't strictly legal. You know, there's
- 18:42the subset of cases where people can bring defamation
- 18:45or false site claims or other torts of that nature.
- 18:48And then there's sort of the stuff in the middle
- 18:49where people are upset. And things can still
- 18:51be very expensive for an artist, even for, you
- 18:54know, what we might say is a frivolous claim.
- 18:58You know or something that lacks legal merit
- 19:00or whatever if it draws enough I or it could
- 19:02still be expensive and chill speech in a way
- 19:05that Filmmakers and other people who make documentaries
- 19:07don't really want so I think it's definitely
- 19:09critical to emphasize, look, here's your legal
- 19:13exposure, here's your audience exposure, here's
- 19:16your industry exposure, when vetting a project
- 19:19on the front end so that the people making the
- 19:21film can have the full tools they need to make
- 19:24the choices they want. And different filmmakers
- 19:25are gonna come out in that calculation differently,
- 19:28but I think our primary obligation as lawyers
- 19:30is not just to spot the legal risk, but also
- 19:32the other things that could impact the project
- 19:34down the line. It's a great answer. And obviously,
- 19:37you obviously handle work on the front end with
- 19:39advising productions too. And so I imagine part
- 19:42of that conversation is going on the entire time
- 19:45in terms of, hey, here's a potential way to mitigate
- 19:47what your risk is going to look like. Same traditional
- 19:50conversation that's always been happening. So
- 19:52let's jump into something that Chris alluded
- 19:55to and more some of these nuts and bolts issues
- 19:58that I think lawyers are dealing with on the
- 20:00vetting side today. Chris, you mentioned editing.
- 20:04And this is one area that I think AI has really
- 20:07affected. We think about the editing process.
- 20:12There's an interesting dynamic, I think, that
- 20:14exists, particularly for documentaries, because
- 20:16for folks who come from potentially the scripted
- 20:19side of the business, they're working with platforms.
- 20:21This very sort of highly interactive process
- 20:23is happening with buyers and distributors, where
- 20:25they're talking about, what are you doing? How
- 20:27are you using these tools? Are we comfortable
- 20:29with it? I think a lot of times, though, for
- 20:31documentary filmmakers, they're working with
- 20:34their lawyers. They're putting something together.
- 20:35They deliver a cut, right? And you're in a different
- 20:38position, I think, when it comes to what you're
- 20:40putting out in the world. So what are the questions
- 20:44you're starting to ask yourself, or maybe you're
- 20:46asking your clients when a cut comes in, as you
- 20:49think about how the editing process is changing?
- 20:52Yeah, well, first of all, I'm asking whether
- 20:54they're using generative AI programs, right?
- 20:57And I think that's a lesson to be learned because
- 21:01I think at first we just assumed people would
- 21:03tell us and then we realized, okay, we have to
- 21:05be very direct with this question. Okay, you're
- 21:08using a particular program. Let's take a look
- 21:11at the terms of service, right? Because not all
- 21:13gen AI programs are created equal, right? And
- 21:17on the one hand, you might have a program that
- 21:20takes anything and everything in the sense that
- 21:23it's trained on the entire internet and whatever
- 21:26comes out as a result of that. On the other hand,
- 21:30you have much more closed systems where the GenAI
- 21:36program can even represent and warrant that everything
- 21:39that it's trained on and everything that creates
- 21:41the output is under license or is owned by the
- 21:45program. And therefore, it's not unlike licensing
- 21:48any non -AI material in the sense that you're
- 21:51going to get reps and warranties that Cover ownership
- 21:55and your right to use it in your film and they'll
- 21:58even indemnify you. So looking at those terms
- 22:00of service is really important as a practitioner.
- 22:05In terms of editing, I think it goes back to
- 22:08what we were talking about before, which is.
- 22:11You know, I think documentary filmmakers are
- 22:14doing the same thing that they've always done,
- 22:16but now they have these amazing tools that allows
- 22:19them to short circuit something or allows them
- 22:21to do something on turbo. An example that we
- 22:26talked about in our pre -meeting was what's called
- 22:30Frankenbiting. And so for people that don't understand
- 22:33or know that term, Frankenbiting is something
- 22:38that documentary filmmakers have been doing for
- 22:41a very long time. And it involves splicing documentary
- 22:46interviews or any kind of documentary audio.
- 22:51and grabbing certain words or deleting them or
- 22:54editing them so that the filmmaker can get a
- 22:57clear statement from the subject. So an example
- 23:00might be, let's say I'm a filmmaker and I interviewed
- 23:04David for eight hours, right? And you get the
- 23:08one sentence that I wanted you to say, except
- 23:11you didn't get one word right, right? So I have
- 23:14to look through your eight hours of interviews.
- 23:17I have to find the word that you actually said
- 23:19correctly. And then I splice it in. So that's
- 23:22frank inviting. And people have been doing that
- 23:24for decades. The difference now is that with
- 23:29AI, I can put your eight hour interview into
- 23:32a program and I can get you to say whatever I
- 23:35want. Right. So filmmakers are still dealing
- 23:39with the same ethical questions. Right. But it's
- 23:42just easier to get the exact result that you
- 23:46want in a way that's maybe not transparent. It's
- 23:50not easy to spot anymore. And the lawyers have
- 23:56to know this. And we can't necessarily identify
- 23:59it when we see it. So again, a lot of these issues
- 24:04are not situations where there's a legal remedy.
- 24:08Maybe you say something that's not defamatory,
- 24:11doesn't place you in a false light. It doesn't
- 24:13really change the nature of what you said. But
- 24:16is it ethically OK? I would say sometimes it
- 24:20is and sometimes it's not, but there's no one
- 24:23clear standard and that's the issue. We've already
- 24:26seen a bunch of litigation relating to this.
- 24:30Our president brought a claim that may or may
- 24:33not have been. frankenbiting done by AI, but
- 24:37it was certainly a frankenbite claim against
- 24:41CBS relating to Kamala Harris's interview. He's
- 24:45now got one going against the BBC for the way
- 24:48they edited one of his speeches. I know that
- 24:52there have been a number of other instances where
- 24:54platforms, including PBS, have refused to run
- 24:58a film because of complaints that came in from
- 25:01outside parties about the language of interviews
- 25:05having been edited inappropriately. I'm not suggesting
- 25:09in any of those instances there was anything
- 25:11inappropriate, but just it opens the door to
- 25:15claims whether or not there was really inappropriate
- 25:18editing. And we're starting to see some organizations
- 25:23come up with solutions for this to try to maintain
- 25:26credibility. My colleagues at Frontline have
- 25:30been doing what they call the transparency project.
- 25:32We publish the transcripts of most of our interviews,
- 25:37the people who appear in the films. We often
- 25:40publish the entire video of the interviews subject
- 25:45to mild editing for either defamation concerns
- 25:49or some other kind of issues that might come
- 25:53up. And it's very popular with audiences. It's
- 26:00very interesting. And I guess the question I
- 26:02would have is, from the litigation perspective,
- 26:05does that help immunize you from liability in
- 26:08the sense that, oh, there's one thing out there
- 26:10that people are viewing that's been mass distributed,
- 26:12and it's in the news cycle. And then maybe there's
- 26:14another stage you have to go through. It seems
- 26:16like a. a good step towards solving the problem,
- 26:18but it's an interesting distinction. It was an
- 26:22interesting risk, and I will tell you, we had
- 26:24a lot of conversation about it because we know
- 26:27that there are people out there that are poring
- 26:29over these transcripts and these videos and trying
- 26:32to compare it to the film and trying to point
- 26:35to discrepancies. For me, At Frontline, it's
- 26:40given me a great deal of comfort because our
- 26:43transparency makes the filmmakers and their editors
- 26:46very alert and very careful about what they do
- 26:49in the editing process. But when it was first
- 26:54proposed, I'll admit that I was a little nervous
- 26:56about it. It kind of goes back to this core concept
- 27:00we've been talking about, which is transparency,
- 27:02trust, right, and accountability in terms of
- 27:05being able to go back to the source materials.
- 27:09I guess, Amanda, question for you. If we really
- 27:13sort of drill down to the most practical level
- 27:16for our audience, for the practitioners who are
- 27:18out there, what are some of the practical steps
- 27:22that you all are taking in your practice? And
- 27:24you guys can jump in on this as well in terms
- 27:26of how you're advising clients, what the process
- 27:29is, how you're protecting yourself. What are
- 27:31some best practices to make sure folks are aware
- 27:34of this and you're catching things as best as
- 27:35you can? Yeah, absolutely. Well, to pick up on
- 27:39something Chris said, I think the first step
- 27:41is to have a full inventory of what we're dealing
- 27:43with. What's been edited? What's been Frankenbitten,
- 27:47if I may invent a verb here? What are we dealing
- 27:50with? And I think this sort of shadow AI use
- 27:54or whatever you want to call it is a real risk
- 27:57and something that we want to make sure that
- 27:58clients are really comfortable disclosing because
- 28:00we can only protect them to the extent we know
- 28:03the full landscape of what we're dealing. So
- 28:06I would recommend that as a first step. It's
- 28:08really important to know what you're dealing
- 28:09with. And then, you know, it's very case by case.
- 28:12I think, you know, as Chris mentioned, too, some
- 28:15cases of editing are going to be okay. Some are
- 28:17going to be riskier than others. You know, one
- 28:19thing, you know, I primarily represent media
- 28:21companies. And one thing people on our side of
- 28:23the bar are aware of is we don't want to be so
- 28:26cautious as to sort of chip away at the First
- 28:29Amendment. you know, overall, and then create
- 28:31an industry standard where maybe the floor is
- 28:34a little bit higher than it needs to be. That
- 28:35said, we also don't want to expose our clients
- 28:37to risk either, you know. So there are a bunch
- 28:40of considerations in mind, but yeah, I think
- 28:42having the inventory and establishing the rapport
- 28:45with the client that, you know, it's best for
- 28:47us to know so we can protect you are the key
- 28:50practical steps I would take. I guess I would
- 28:54add that fact checking is absolutely critical.
- 28:59And in many, many instances, one of the first
- 29:03pieces of advice that I give to independent filmmakers
- 29:06who come to us at the clinic is that they should
- 29:10have somebody who was not part of the creation
- 29:13of the story, not one of the advocates for the
- 29:16story, who's actually doing some fact checking.
- 29:19I mean, I come from a journalism background.
- 29:21I worked in newspapers for a long time. At the
- 29:24Chicago Tribune, the old saying was, if your
- 29:26mother says she loves you, check it out. It's
- 29:32funny, but it also has got a real ring of truth
- 29:36about how to best protect yourself against lawsuits.
- 29:42are there to report against your own biases who
- 29:45are checking the facts really carefully, like
- 29:48the New Yorker magazine was famous for and still
- 29:51is famous for. In a potentially controversial
- 29:55film, that's one of the most important best practices
- 29:57I can think of. Yeah, and I would just add, just
- 30:01to go back to transparency, I think it's more
- 30:05than transparency with the film team and with
- 30:08counsel. I think filmmakers need to be transparent
- 30:12with other stakeholders, like distributors and
- 30:15broadcasters, and in particular with insurance
- 30:19companies. I think the worst case scenario would
- 30:23be you get a claim on some AI generated material
- 30:27saying that this is infringement or it violates
- 30:31somebody's right of publicity or violates one
- 30:32of the deep fake laws and you're not covered
- 30:36by insurance because you either didn't disclose
- 30:38it, or you answered a question on that insurance
- 30:43application falsely because you used a program
- 30:49that didn't get the requisite representations
- 30:53and warranties. It's interesting because I talked
- 30:56to an underwriter maybe two or three years ago,
- 31:00and I asked them the question, what do you think
- 31:03about the use of AI materials and Are you going
- 31:06to ensure them if you get claims on them? And
- 31:10the answer was, we'll ensure them as long as
- 31:14clients are engaging in AI use responsibly. So
- 31:19I said, well, thanks for the non -answer. But
- 31:21that's not all that helpful. But I think years
- 31:24on, I think things have changed a little bit.
- 31:26And as I mentioned before, Responsible AI use
- 31:30to me means disclosure with your insurance company.
- 31:33If you're not getting the reps and warranties
- 31:35that would be nice to get from a closed system
- 31:39and you're using a more open system, at least
- 31:41explain the program that you're using to the
- 31:44insurance company so that if you get a claim,
- 31:47they can't wriggle out of coverage. It's a great
- 31:51point, and it goes to a question that's in the
- 31:52chat here that I was meaning to get to. Professor
- 31:54Jay Dougherty from UCLA, who joins many of our
- 31:57seminars, as many of you know, was asking specifically,
- 32:00do these contractors between the producers and
- 32:02creative contributors, or the distributors and
- 32:05financiers, include provisions that require these
- 32:07types of disclosures, including disclosures around
- 32:11things like script annotation, he mentions. And
- 32:13he also asked about E &O clearance procedures.
- 32:16And I guess part of what I hear you saying is,
- 32:19in part, the answer is yes. But even if they're
- 32:22not necessarily required as a practice, it sounds
- 32:25like you're volunteering this information. How
- 32:27do you advise clients in terms of that disclosure?
- 32:30Yeah, it's a great question. I'll start with
- 32:32insurance one first. I mean, E &O insurance applications,
- 32:36if anybody's ever looked at one, there's no one
- 32:39size fits all application. And half those questions
- 32:42are almost impossible to answer for a given film
- 32:45because it just doesn't apply. And so very often,
- 32:49rather than checking a yes or a no or an NA,
- 32:53you're having to supplement it with an explanation
- 32:56for why this question isn't quite a yes or no
- 32:58answer. So even though most of these applications
- 33:03have not been updated in years, and they're not
- 33:06asking about AI, and for the most part, these
- 33:09clearance procedures are not addressing AI use,
- 33:12I still think disclosure is the way to go, particularly
- 33:14in a world where there is a lot of litigation
- 33:18around these issues. And we don't know where
- 33:21claims are going to come from. So better to be
- 33:24transparent than to hide it. As far as contracts
- 33:29go, we are drafting contracts that address the
- 33:32use of AI. So in most cases, when particular
- 33:36crew are being hired, there is a prohibition
- 33:39against the US use of AI unless it is expressly
- 33:43given permission by the producers of the film.
- 33:47Because again, you just want to know exactly
- 33:49what's going on. Great. So then if we dig just
- 33:55a little deeper into that conversation or that
- 33:57subject, we had talked about record keeping earlier.
- 34:01It's kind of a pedestrian subject for most folks
- 34:03out there. But for lawyers, this is the good
- 34:05stuff. So what records are you really specifically
- 34:09advising your clients to keep along the way that
- 34:12are the most useful from a legal and risk management
- 34:14perspective in anticipation potentially of getting
- 34:17claims or having to deal with these issues? It's
- 34:20another one of these situations where creating
- 34:23a log is probably really valuable and important.
- 34:27A log that will record what programs or platforms
- 34:31were being used, how they were used with respect
- 34:36to which particular pieces of content, which
- 34:40prompts were used and how the prompts were entered
- 34:43as well. My preference right now, quite frequently
- 34:48is to tell clients to prepare this in a setting
- 34:52that could be privileged or work product, working
- 34:55with counsel. But in many instances, that's not
- 34:59always going to be possible. the more information
- 35:04you have about how something was created, generally
- 35:07speaking, again, assuming ethical behavior and
- 35:11reasonable behavior. Back to Chris's point about
- 35:13what the insurance company said, having documentation
- 35:17is going to be helpful in dealing with potential
- 35:20problems down the road. Again, back to the Archival
- 35:24Producers Alliance, I think they've got some
- 35:26really good guidelines on record keeping and
- 35:29what should be recorded and how it can be recorded
- 35:31in the way that that's really clear and usable
- 35:34after the fact. And so I urge people to go take
- 35:39a look at that if they're interested in the record
- 35:41keeping question. Yeah, one thing that comes
- 35:44to my mind that Chris mentioned previously is
- 35:46the the license terms and the user terms for
- 35:51these platforms as well. I imagine that can go
- 35:53a long way, Amanda, to showing you've done your
- 35:55homework, at least when you're getting these
- 35:57types of claims. Yeah, absolutely. And I mean,
- 36:00it's also useful, too, for copyrightability,
- 36:02which I know we're at. we're going to get to.
- 36:04But showing your homework is absolutely the best
- 36:07practice, just as a matter of record keeping.
- 36:09And then, yeah, just to foreshadow our copywriting
- 36:11discussion as well. Yeah, well, let's jump into
- 36:13that. So a new topic here, obviously copyrightability.
- 36:17It's familiar to lawyers. There's a new conversation
- 36:19that AI has. God, I'm going to say that AI has
- 36:23prompted, to use a terrible pun here, when we're
- 36:26talking about what is copyrightable. And the
- 36:29creative folks, the filmmakers are finding themselves
- 36:32in this new conversation where they're talking
- 36:33to the lawyers about, how do I protect my work
- 36:36if I'm using AI in it? If I use AI to do X, Y,
- 36:39or Z, can I still get a copyright? Before we
- 36:42get into the practical tips and maybe the prompts
- 36:44and some of the other elements, Amanda. What's
- 36:47the big picture you're thinking about here for
- 36:49your clients when we talk about copyrightability,
- 36:51specifically in the context of documentaries?
- 36:53Yeah, absolutely. The big picture is that it's
- 36:56changing and it's very fluid. We have a couple
- 36:59of very, very few number of cases that are giving
- 37:01us some guidance posts, but the whole thing is
- 37:04very unsettled. So for example, as it stands
- 37:06right now, it seems like purely generated AI
- 37:10works that might be used in the documentary are
- 37:12not going to be copyrightable. That comes from
- 37:14the Fowler case, which was decided at the circuit
- 37:16level, and then the Supreme Court declined to
- 37:19hear it. So that's sort of where things stand
- 37:21right now. cases, for example, where somebody
- 37:25inputted, you know, scores of prompt and created
- 37:29a comic book and the copyright office found that
- 37:32the prompts were copyrightable, not all of the
- 37:34output was. We have cases in Colorado and also
- 37:38recently filed here in the central district that
- 37:40have sort of more extensive records of human
- 37:44assisted AI generation. We have One of my favorite
- 37:50examples is a piece called A Single Piece of
- 37:52American Cheese or something to that effect,
- 37:55don't quote me on the exact title, but there
- 37:58the people seeking copyright had extensive records
- 38:00of how that particular piece was created and
- 38:03they were able to get a copyright in their work.
- 38:06But, you know, right now, I mean, it's still
- 38:08very much a practical concern for filmmakers,
- 38:11and filmmakers should think carefully about how
- 38:13to put as much of a human fingerprint as possible
- 38:16on AI -assisted work. These questions raise chain
- 38:21of title implications, so it's just good to be
- 38:24aware of and to keep an eye on as well, because
- 38:26this doctrine is evolving quickly. Yeah, and
- 38:30so in the context of these documentary films,
- 38:33obviously, To me, it's so interesting because
- 38:36there's so many pieces to what can exist within
- 38:39the context of a documentary. Oftentimes, in
- 38:41the context, let's say, of a scripted series
- 38:43or a film, there's this idea of where does the
- 38:45idea come from? How are the characters being
- 38:47generated? But it's kind of a different conversation
- 38:50around a documentary where there's the interviews,
- 38:53there's the human element to it, there's a storytelling
- 38:55quality to it, and then there's using AI to amplify,
- 38:59I imagine, all of those elements. And I think
- 39:02that The question on everybody's lips is sort
- 39:05of, where do you draw the line, right? What's
- 39:07enough human influence in order to get copyright
- 39:11protection? And how do you get over that barrier?
- 39:15I guess I'll ask you, Chris, because this is
- 39:17a tough one, especially. Why don't you jump in
- 39:19here? It's on everybody's lips, this question.
- 39:23How do you talk to clients about how to get over
- 39:25that barrier? Well, what I would say is if you're
- 39:27asking that question, you're probably in a dangerous.
- 39:29place to begin with, right? If the question begins,
- 39:35well, I had an idea, and AI took me the rest
- 39:42of the way. It wrote the script. It generated
- 39:44the video. I don't think there's a right answer,
- 39:48especially at this moment in time. But I totally
- 39:51agree with Amanda that it's very much, we just
- 39:53don't have enough cases to know where the line
- 39:56is. But I think it's disclosing the risks of
- 40:00not having copyright protection if the work just
- 40:04isn't human enough. And then what is the knock
- 40:08on effect of that? Well, you might have a chain
- 40:10of title defect. You could open yourself up to
- 40:12piracy. You know, it might be difficult to get
- 40:16a copyright registration. So to protect your
- 40:19your work in court, the federal court would be
- 40:21very difficult. You know, when you're registering
- 40:26a documentary for copyright, for example, you
- 40:29already have to grapple with the questions of,
- 40:31okay, what am I actually protecting in this film
- 40:34and what material needs to be excluded, right?
- 40:37So for a typical documentary where you have archival
- 40:39material that you don't own, you have to exclude
- 40:42that material from the copyright registration.
- 40:46You're going to have to do the same with AI created
- 40:48works. And that's really where all of these cases
- 40:51started, where registrations with the copyright
- 40:55office gone awry, right? Some of them went through,
- 40:58I forgot which case it was where the registration
- 41:01actually went through, but because there was
- 41:03so much media coverage around it, the copyright
- 41:05office kind of pulled it back and said, well,
- 41:08hold on a second, you lied on your application.
- 41:11So I won't be surprised if there's more cases
- 41:13like that, but yeah, there's certainly a lot
- 41:17of interesting movement in this area. Well, and
- 41:21I'll just say too for the wonk some of us we
- 41:23had a really interesting chat a couple weeks
- 41:24ago with Professor Edward Lee at Santa Clara
- 41:26who runs chat GBT is eating the world. And he's
- 41:29come up with kind of a novel way to think about
- 41:31copyright using AI in the vibe coding context
- 41:34that people may apply here it's the selection
- 41:37and arrangement argument, and very very broadly,
- 41:40it goes that by inputting enough prompts your.
- 41:44adding sufficient human creativity to get a thin
- 41:46layer of copyright protection. I mean, that's
- 41:48super novel. It's not been tested. I just thought
- 41:51I'd throw it in there in case we had other copyright
- 41:54dorks among us. So this one seems like it is
- 41:57going to develop, though. It's hard to imagine
- 42:00given. the likely uses we're going to see from
- 42:05big studios and everyone else of AI that the
- 42:08law is going to stay static in the current form.
- 42:14But hard to tell whether that will come from
- 42:16the courts or from Congress, if laws ever come
- 42:20from Congress again. But it's difficult to imagine
- 42:24that things will stay the way they are on this
- 42:26subject. One thing that I would, sorry David,
- 42:29just one thing I wanted to add there is that
- 42:31I think as practitioners, we ought to be careful
- 42:33of a situation where, let's say a client says,
- 42:37well, you know, I only did two prompts and I
- 42:40got the material versus, oh, I actually did 600
- 42:43prompts and I got to the material. That's an
- 42:46important distinction, but it's certainly not
- 42:48dispositive, right? Because we know under copyright
- 42:51law, you need something tangible, you need something
- 42:53creative. So even if it, you do something 600
- 42:57times, it doesn't automatically become creative
- 43:00from a human authorship standpoint. So even if
- 43:05there's a lot of so -called work on the human
- 43:07side, it often will not rise to the level of
- 43:10copyright protection. And so what sort of practical
- 43:13advice are you giving your clients about sort
- 43:15of how to use these tools? I mean, Dale, you
- 43:17talked about keeping track of prompts. You must
- 43:21be in this conversation constantly about where
- 43:23it might be a good idea and where maybe you should
- 43:24stay away from it. I wouldn't say I'm in it constantly,
- 43:28but it definitely comes up frequently for all
- 43:31of us. I think the answer is, in most cases,
- 43:35I tell clients that if they're going to be creating
- 43:37something using AI, that they are unlikely to
- 43:41be able to count on it being copyrighted. That's
- 43:44something for them to think about going into
- 43:46it. As Chris said, there's only so much you can
- 43:50do to try to create, whether it's by arrangement
- 43:55of the prompts or by some other theory to protect
- 44:00this work that you're going to be creating. So
- 44:04most of the filmmakers that I've dealt with,
- 44:07at least, have accepted the fact that if they're
- 44:10creating something that they may have to live
- 44:12with people using it and violating what they
- 44:16think of as their proprietary rights. It's interesting
- 44:20and you mentioned this conversation is ongoing.
- 44:23One aspect of the conversation that we haven't
- 44:25really touched upon, and if we step back for
- 44:28a moment, is the broader conversation about Should
- 44:32AI be used in these types of works or not, right?
- 44:35This is one of these issues where Hollywood is
- 44:37sort of really, and I imagine the news media
- 44:40as well, really in conflict with itself and all
- 44:42the way from the optimists and the early adopters
- 44:44to the do -mers. And there can also be a cultural
- 44:48element to not wanting to take a side one way
- 44:50or the other. Are you seeing this play out within
- 44:53the documentary community as well? Sure. And
- 44:57in fact, a lot of people just instinctively are
- 45:00anti -AI. because of all the press about the
- 45:04fact that it's going to cost jobs, that it's
- 45:06going to replace humans doing any number of different
- 45:09kinds of jobs. But we've also seen some really
- 45:12good uses of AI that are really terrific. The
- 45:16first one that most of us were exposed to at
- 45:19Sundance years ago was a film called Welcome
- 45:22to Chechnya, where to protect the identity of
- 45:26people who might have been subject to prison
- 45:28or worse in Chechnya. because that they were
- 45:31gay. The faces were reconstructed. They weren't
- 45:35just blurred. They were reconstructed using AI.
- 45:39That's a terrific use. It was fully disclosed.
- 45:42And yet there were still people who were opposed
- 45:45to it and who were critical of it because they
- 45:47just are instinctively anti AI. We've seen it
- 45:50with voice. recreations, the Anthony Bourdain
- 45:54example that caused all sorts of controversy,
- 45:58the Andy Warhol's diaries, as long as it's fully
- 46:03disclosed and it's not done in a way that deceives
- 46:07anyone or steals rights from them, it seems like
- 46:11there are great uses for AI. going forward, including
- 46:16for recreations. I mean, recreations have been
- 46:19part of documentaries going all the way back.
- 46:21Nanook of the North, one of the most famous early
- 46:23films, had recreations in it. As long as you're
- 46:27not deceiving people and taking something away
- 46:30from them, I think using AI is OK. But the repercussions
- 46:36for human workers are really serious, and it
- 46:41has caused a lot of the blowback. One of the
- 46:46things that I guess comes to mind when you mention
- 46:48that is a very current topic right now in the
- 46:50news. Maybe it's documentary adjacent, but it's
- 46:54the film that was recently entered into, I believe,
- 46:56the Tribeca Film Festival, recreating on a sort
- 46:58of more of a scripted version of protests that
- 47:04occurred in Iran, right? And it goes to what
- 47:06you were talking about, which is it's probably
- 47:09a space where, obviously, documentary filmmakers
- 47:11are not able to get in there to tell the story.
- 47:14But it's a fascinating conversation to have with
- 47:17the clients. So let's jump into. Another easy
- 47:22subject here, now that we've tackled copyright
- 47:24ability, let's talk about fair use. For lawyers
- 47:29who work on documentaries, this has always been
- 47:33a fundamental aspect of practice where you've
- 47:36added critical value, but AI is once again changing
- 47:39the conversation around this across the media
- 47:42spectrum. Chris, you got into this when we talked
- 47:45about the input versus outputs distinction a
- 47:47little bit. Maybe you could set the table for
- 47:51us on what the fair use conversation looks like
- 47:54with documentary filmmakers these days with these
- 47:57tools now being available. Oh, you're muted.
- 48:06Chris, I think you're muted. Sorry about that.
- 48:10Yeah, I was going to say, hate to say it, but
- 48:12again, a moving target, right? Because we are
- 48:15working with a Fast evolving technology and also
- 48:19a lot of litigation. At the district court level
- 48:23that will soon go to appellate courts and so
- 48:26we're going to learn a lot more. I think in the
- 48:27next couple of years. But as you mentioned, there's
- 48:30really 2 central questions when we're talking
- 48:32about fair use 1 is the. Input or the training
- 48:36question, right? Is the. Training of these on
- 48:40unlicensed material going to be fair use and
- 48:43there's a few cases on that. And then the second
- 48:46question, which I think is actually more relevant
- 48:48and easier for filmmakers to get their heads
- 48:51around, which is, is the output going to be fair
- 48:54use, right? So regardless of how the LLM that
- 48:57you're using is going to be, regardless of whether
- 49:01that's infringement or fair use, is your material
- 49:04that comes out substantially similar to a copyrighted
- 49:10work that you didn't license. And actually starting
- 49:12with the second question first, I don't think
- 49:15much has changed in that regard. I think we're
- 49:19still asking, and this is reiterated with the
- 49:24Warhol Supreme Court case, it really comes down
- 49:27to transformativeness. Is the purpose and character
- 49:31of your use transformative? from the original
- 49:36purpose and character of the use. But I guess
- 49:39even before that, you have to ask whether there
- 49:42is substantial similarity to begin with, whether
- 49:44you had access to the material. And if you have
- 49:48a bunch of prompts that say, well, listen, I
- 49:51want an artwork in the form of Warhol and this
- 49:55particular artwork, well, that's a red flag,
- 49:58right? As a practitioner, you want to identify
- 50:00those things and say, well, not only does this
- 50:03look like it, but you actually asked it to generate
- 50:05something that was exactly like that protected
- 50:08work. And this, again, another good reason to
- 50:11have good record keeping. So I think on the fair
- 50:16use output question, not much has changed, although
- 50:20there's an important case that survived a motion
- 50:24to dismiss that'll be really interesting to track.
- 50:27This is the New York Times versus OpenAI case,
- 50:30where The New York Times is saying well listen,
- 50:33ChatGPT can sometimes reproduce whole portions
- 50:38of our articles that look pretty close to the
- 50:40original articles. So if they're able to establish
- 50:45that that's the case, they can establish market
- 50:48harm pretty easily under the fourth factor of
- 50:50fair use because people can just go to ChatGPT
- 50:53to read the news rather than New York Times getting
- 50:59subscribers to read the same thing. It'll be
- 51:02really interesting to see how that case evolves,
- 51:03of course, because with those New York Times
- 51:06articles, we're dealing primarily with facts
- 51:09and theories and historical interpretations of
- 51:11facts, all of which is not copyrightable. But
- 51:14the expression of that as a whole is, and I have
- 51:17looked at some examples of some of the excerpts
- 51:21that ChatGPT generated next to New York Times
- 51:25articles, and it's actually quite fascinating.
- 51:28It's very, very close. I would not have thought
- 51:31that ChatGPT would generate something that close.
- 51:34So we're tracking that case. That's an important
- 51:37output case that we'll want to see where it goes.
- 51:43In terms of the input, there's the Barts case,
- 51:46there's the Cadry case, and in both cases, they
- 51:49found that the training of an LLM on unlicensed
- 51:56material was transformative enough for it to
- 51:58be fair use. But they've also been very, very
- 52:01careful in those cases to say this is a very
- 52:04limited decision. And if we were talking about
- 52:09the reproduction of works as a whole where the
- 52:12output was substantially similar to the underlying
- 52:17work, our opinion would probably be different.
- 52:20And the issues in those cases is that there was
- 52:22no evidence of that. It was really limited to
- 52:25the input question. Let me ask, how does that
- 52:28translate into a documentary filmmaker getting
- 52:29into trouble when using tools like this? Or you're
- 52:32reviewing a cut and you're looking at something
- 52:35that's been created with AI and you're concerned
- 52:37that there's a potential substantial similarity
- 52:39issue. Yeah, well, the first question is, which
- 52:43program did you use? And again, returning to
- 52:46the question of, is it one of these open systems
- 52:49where it's the Wild West, it could have drawn
- 52:52from anything, right? And so we're looking at
- 52:54the prompts to see what you were trying to create,
- 52:56how much risk is there in terms of directing
- 52:59the LLM to produce a specific result. At the
- 53:02other end of the spectrum, you have archives
- 53:05that are using, they have proprietary LLMs that
- 53:10are working on a closed system using only licensed
- 53:13or owned works. And I can imagine how that might
- 53:17be limiting creatively, right? Because you're
- 53:19working with less works, but at the same time,
- 53:21from a risk perspective, it's so much lower because
- 53:24you're getting the reps and warranties that you
- 53:26need from the licensor. In most cases, they'll
- 53:30even indemnify you from claims, which is great.
- 53:33And then when it comes to insurance, it should
- 53:35be an easy thing to cover because it's just like
- 53:38any other archival license in the sense that
- 53:40you're getting those reps and warranties in the
- 53:41indemnification. With the open systems, they
- 53:44often ask you to indemnify them in the event
- 53:48of a claim. So it's pretty much the exact opposite
- 53:51situation. Yeah, and of course, there's the question
- 53:53of ownership of those materials as well, depending
- 53:55upon the licenses, terms that are in place. Amanda,
- 53:58jump in here. What are you thinking in terms
- 54:00of how you're seeing your filmmakers work with
- 54:04these tools and what a fair use kind of analysis
- 54:07is coming up in your conversation with them?
- 54:09Yeah, absolutely. Well, I'm definitely tracking
- 54:11the New York Times versus opening up the case.
- 54:12That'll be interesting. I think several of the
- 54:15studios have sued the AI platform, made Journey
- 54:182, which, just to set the table they're alleging
- 54:21that mid journey can produce very near replicas
- 54:25of their intellectual property. So beloved Disney
- 54:27icons, for example, things like that. And I think
- 54:29that'll be a really interesting test of the fair
- 54:32use doctrine in the output context as well. Yeah,
- 54:36I mean that was my primary ad. I think this is
- 54:39unsettled. I think as Chris said, substantial
- 54:41similarity kind of remains the grounding point.
- 54:44If the output is substantially similar to a copyrighted
- 54:48piece of material, I think you're gonna be on
- 54:49some fairly shaky ground. And yeah, I think it's
- 54:53definitely something we as predictioners need
- 54:55to flag. Dale, do you want to jump in here and
- 54:58add anything to the discussion? I don't know
- 55:00that I have much to add. I think that's a really
- 55:02smart analysis from both Amanda and Chris on
- 55:06this front. We haven't talked much about right
- 55:09of publicity. That's another place where there's
- 55:12a really strong potential for claims when AI
- 55:17is used. It's a little less likely to be inadvertent.
- 55:22Some of what we're talking about, especially
- 55:24with the open systems, are inadvertent situations
- 55:28where the filmmakers might not be intending to
- 55:31copy. It just happens. In the right of publicity
- 55:35context, it would be a little more intentional.
- 55:38Again, one of the key issues from my perspective
- 55:42is always, you know, to to be really transparent
- 55:46and ethical about things. And if something appears
- 55:50like it's close, that's usually a very intense
- 55:53conversation between the council and the filmmakers.
- 55:58Because the truth is, most of the filmmakers
- 56:01I deal with, at least the independent filmmakers,
- 56:04probably the two of you as well, can't afford
- 56:07the litigation, whether they've got insurance
- 56:09or otherwise. They can't afford the distraction
- 56:12and diversion. They can't afford the impact on
- 56:16their distribution, which may be very severe,
- 56:19and they can't afford even the legal fees up
- 56:22to the time of their deductible being filled.
- 56:26Yeah, and I think big picture. I mean, I think
- 56:28a through line for this entire conversation is
- 56:31that, you know, the folks we're talking about
- 56:33who produce these documentaries. They're often
- 56:36really committed news people. They're committed
- 56:39storytellers. They care deeply about the subject
- 56:42that they're bringing to the world. And I think
- 56:46when they're in this conversation with you, they
- 56:49want to make sure they get it right, right? They
- 56:50respect other people's work as well. Yes. and
- 56:55the kind of work that you folks do, there is
- 56:57sort of a gut element, I think, that underlies
- 57:00all of this in terms of a sense of fairness and
- 57:03what's right. So as we move forward, we've got
- 57:05a few minutes left here. Maybe I could just sort
- 57:07of go around the houses here and ask each of
- 57:10you to sort of share with us a parting thought
- 57:12on a piece of advice you would give to either
- 57:14practitioners or to your clients today as they
- 57:17continue to navigate this. Look, it's a very
- 57:19competitive. landscape, obviously. Things are
- 57:22only getting more complicated, but some things
- 57:25seem to remain the same. Amanda, maybe you could
- 57:28get us started. Yeah, I mean, I'll start with
- 57:30a sort of the big picture thing that animates
- 57:33me, which is our job as I see it is to help our
- 57:36clients tell a story and tell them. keep telling
- 57:40stories if that's what they wish. So that involves
- 57:41minimizing the risk of litigation. That involves
- 57:44all the things we've discussed today. But I think
- 57:46having that as a North Star is really important
- 57:49in thinking through all of these issues, from
- 57:51an ethical standpoint, from a legal standpoint,
- 57:53from a practical standpoint. And otherwise, we're
- 57:56here to empower our clients to tell their story.
- 58:01Fantastic. Dale, maybe you could jump in next.
- 58:04And I'd also mention we have one other question
- 58:06in the chat here. that kind of goes to something
- 58:10you've talked about, which is fact checking.
- 58:13And is AI being used for fact checking experience?
- 58:17And how are you thinking about that moving forward?
- 58:19Really interesting question. And I think I'm
- 58:22glad that it came up. I mean, people are using
- 58:26AI for fact checking and also for defamation
- 58:30and privacy review. And I would say with mixed
- 58:33results at this point. For one thing, everybody
- 58:36knows about hallucinations, so I'm not here to
- 58:40scare everybody about hallucinations. But if
- 58:43you're using AI for fact checking, you still
- 58:45have to fact check your fact checker, because
- 58:48mistakes get made by the AI, just like they get
- 58:51made by people, but I think at a higher rate
- 58:54in many instances. And as we've all seen in the
- 58:58legal context, judges are getting more and more
- 59:01aggressive and are taking more and more severe
- 59:05steps to address the hallucinations that are
- 59:10not being fact -checked by lawyers. I think the
- 59:12same thing will happen in the defamation and
- 59:15privacy context. On the vetting side of things,
- 59:19there have been a few attempts to create vetting
- 59:22software using AI. I've seen the demonstrations
- 59:27of it. I have not actually used it. The problem
- 59:30is that the AI will identify everything that
- 59:33might be a problem, which actually creates more
- 59:37stress and more concern than really needs to
- 59:41be shared with clients or frankly with the lawyers
- 59:44themselves. So I don't think the vetting software
- 59:47is quite there yet. From a fact checking standpoint,
- 59:51as a starting point, it may be useful for checking
- 59:54particular facts. I'm not sure I'd rely on it
- 59:58for an entire documentary film. And again, it
- 1:00:01would have to all be double checked by someone
- 1:00:04human who has real judgment to bear. It's a great
- 1:00:08point. We talked about AI replacement. We didn't
- 1:00:10get into the piece about lawyers worrying about
- 1:00:12their jobs. But I think your point goes to the
- 1:00:15value of human judgment, especially the folks
- 1:00:17like you have been experienced for a long time.
- 1:00:20Chris, maybe you could have the last word. And
- 1:00:22there's a question in here also about going back
- 1:00:26to fair use. And should there be maybe a a more
- 1:00:30specific notion about how fair use should be
- 1:00:32applied in the documentary context. I'm not sure
- 1:00:34if that's something you want to touch, but if
- 1:00:35you had any parting thoughts on that as well.
- 1:00:37You're right, because as an office, we were talking
- 1:00:41about this on Wednesday, the extent to which
- 1:00:43there are cases that state that documentary filmmaking
- 1:00:47is part of one of the stated purposes that automatically
- 1:00:53are fair use. The answer is not really. I mean,
- 1:00:56there are cases that kind of say that, and we
- 1:00:58certainly quote them in our fair use opinion
- 1:01:00letters, but it's not among the stated purposes
- 1:01:03that for which the fair use doctrine was actually
- 1:01:07created under statute. However, it has all of
- 1:01:10the hallmarks, especially with a pretty straightforward
- 1:01:14talking head documentary for something to be
- 1:01:16transformative. So it's almost like a head start.
- 1:01:21unfair use in the documentary context, but certainly
- 1:01:24not automatically. Dale, you wanted to say something?
- 1:01:27Yeah, I don't want to interrupt you, Chris, but
- 1:01:29I would highly recommend to whoever asked the
- 1:01:31question, anyone who's interested, the 10th Circuit's
- 1:01:34recent decision in the white monkey case, which
- 1:01:37really goes through step by step on all four
- 1:01:40of the factors to talk about why documentaries,
- 1:01:43at the very least, are more likely to qualify
- 1:01:46for fair use than than other types of content.
- 1:01:50And I agree with Chris, it's not a presumption,
- 1:01:53although I've argued it's a presumption in the
- 1:01:55past. And I certainly believe it should be a
- 1:01:58presumption for documentary filmmakers. But the
- 1:02:02White Monkey case is super helpful for documentary
- 1:02:05filmmakers and their lawyers. And thank God that
- 1:02:08that decision came down the way it did. Yes.
- 1:02:11So you have a few hours to read that decision.
- 1:02:15It's a long one. It's a long one, but it's all
- 1:02:17good. And it's almost like, which of these great
- 1:02:20quotes am I going to use in my opinion letter?
- 1:02:23So very, very. Yeah, in terms of final words,
- 1:02:28just to kind of piggyback off what Amanda and
- 1:02:30Dale said, I would say, don't be afraid of the
- 1:02:35technology as a practitioner. It's here. You
- 1:02:38have to embrace it in some ways in the sense
- 1:02:41that you have to be comfortable with it. You
- 1:02:43have to understand how it works and how your
- 1:02:45clients are using it. But there should be a human
- 1:02:48at the end of every process without fail. So
- 1:02:52on the vetting side, as practitioners, where
- 1:02:55probably most of us are using some kind of law
- 1:02:58AI tool, you still have to check the work. There's
- 1:03:02a lot of hallucination cases. And even short
- 1:03:05of that, it's maybe not thinking about certain
- 1:03:08things that a human would. So you have to double
- 1:03:10check that work. And for clients, similar thing
- 1:03:16applies. If you're creating something that is
- 1:03:19largely created by AI, I think a human really
- 1:03:23needs to step in and take a look, not just for
- 1:03:25legal issues, but also for questions of audience
- 1:03:28engagement. I had a client that was producing
- 1:03:32a true crime doc. And they decided very early
- 1:03:36on that they wanted to do 100 % of the recreations
- 1:03:39through AI. And they recreated a lot of the archival
- 1:03:43materials through AI. We worked through all of
- 1:03:47the legal issues. But when they put it in front
- 1:03:50of a test audience, it just, it did not work.
- 1:03:54People could tell. People were angry. And so
- 1:03:57they went back to the drawing board and had to
- 1:04:00shoot a lot of recreations. and use real archival
- 1:04:03material. So I think that kind of feedback, whether
- 1:04:06it's legal or questions of whether it's ethical
- 1:04:10or thinking about whether an audience will accept
- 1:04:14it, all questions that they should be asking
- 1:04:17themselves. Great. Fantastic point to close on.
- 1:04:21I think, again, it goes back to some of these
- 1:04:23core principles we've been talking about, trust,
- 1:04:25integrity, the relationship to the audience,
- 1:04:28fidelity to the story. And obviously, your clients
- 1:04:33are very fortunate to have you in their corner
- 1:04:35as they confront all these issues and work through
- 1:04:37them. I want to thank you all for sharing your
- 1:04:40expertise and your insights today as well. It's
- 1:04:42been a real pleasure to have this conversation
- 1:04:44with you. We hope you'll come back and talk to
- 1:04:46us again. Thanks, David. Thank you for having
- 1:04:49us. It was really fun. Thank you to my fellow
- 1:04:51panelists. That was terrific. Same. Thanks, Dale.
- 1:04:54Thanks, Amanda. Thanks, everyone. This was awesome.