The Ethics of AI in Documentaries: Rachel Antell on APA’s Mission to Protect Archival Integrity

Rachel Antell has worked in documentary film production for over two decades but found her true passion in archival producing. In 2014 she co-founded Sub-Basement Archival, where she’s had the honor of being archival producer on dozens of documentaries that have screened worldwide, and garnered FOCAL Awards and an Emmy nomination for research and use of archival footage. In 2023, Rachel co-founded the Archival Producers Alliance (APA) with Jen Petrucelli and Steph Jenkins. 

The APA was created to organize archival producers, educate the film community about the craft, and advocate to make archival materials accessible and affordable to independent filmmakers. The APA's first initiative addresses Generative AI’s impact on documentary film and archival work. Last year, the APA published the first industry-wide guidelines, Best Practices for Use of Generative AI in Documentaries, with endorsements from over 50 organizations, foundations, production companies, and film festivals, as well as dozens of filmmakers.

Transcript

SPEAKER_01 1:15

I'm Carol Dean, president of the Heart Productions. Join us to learn new guidelines for AI to prevent false historical information. This is important.

SPEAKER_00 1:29

Our very special guest today is Rachel Antell. She's worked in documentary film production for over two decades, but found her true passion in archival producing. In 2014, she co-founded Sub Basement Archival, where she's had the honor of being archival producer on dozens of documentaries that have screened worldwide and garnered focal awards and an Emmy nomination for research and use of archival footage. In 2023, Rachel co-founded the Archival Producers Alliance, also known as APA, with Jen Petricelli and Steph Jenkins. The APA was created to organize archival producers, educate the film community about the craft, and advocate to make archival materials accessible and affordable to independent filmmakers. The APA's first initiative addresses generative AI's impact on documentary film and archival work. Last year, the APA published the first industry-wide guidelines for best practices for use of generative AI in documentaries, with endorsements from over 50 organizations, foundations, production companies, and film festivals, as well as dozens of filmmakers. And Carol, this is very important information for documentary filmmakers.

SPEAKER_01 3:11

Yes, it is. It's a new world, and we have to keep up. But thank you very much, Rachel, for joining us.

SPEAKER_02 3:18

Yeah, thanks so much for having me.

SPEAKER_01 3:20

So I'm uh I want to understand the Archival Producers Alliance, its mission, and perhaps an overview uh of your mission in the documentary industry.

SPEAKER_02 3:34

Yeah, uh, that's a great question. And for people who don't know um what archival producers do, our work involves researching, licensing, and also verifying the audiovisual um archival materials that are used in all kinds of media, but primarily in non-in um documentary films. And so in the spring of 2023, my partner and I started seeing synthetic historic images uh being brought into documentaries, and they were indistinguishable from the real historical images that we were bringing in. So, given our role as like fact-checkers of third-party media, this raised big questions and concerns for us, um, both about how this wouldn't somehow muddy the historical record, and then also how documentary as a form would maintain its commitment to historical truth and the trust that it had built with audiences. Um, so my colleagues, Jen Petrosali Steph Jenkins, and I gathered a group of archival producers from around the country to discuss how generative AI could impact uh archival work as well as the documentary field more broadly, and whether there was anything we could do to move the needle. And that led us to collectively writing and releasing an open letter that outlined the concerns and was printed in full by the Hollywood Reporter. It was signed by over a hundred practitioners. And through the process, we found that people across the documentary industry really were hungry for guidance. They knew that this technology was, you know, coming fast and furious into their world and wanted to understand how to use it responsibly, if they were how to consider whether they wanted to use it and then if they did, how to use it responsibly. So we formalized our group as the Archival Producers Alliance or APA. And then in conversation with people across the documentary field, as well as lawyers and scholars and technologists, um, we we wrote this set of best practices that are really born of the values that the documentary community has always held. Um, and that was that was really the goal. Um, I mean, because we knew that within the tech field, there are a lot of other people who are doing work around quote-unquote ethics of AI and creating their own standards around it, and that's great, but we didn't believe that their primary interest was necessarily the protection of the documentary field. So our work was very intentionally grassroots because we felt that any set of standards, for it to be embraced by the documentary community, it really needed to come from the community itself.

SPEAKER_01 6:13

Absolutely. From the Heart has given many grants to documentary films. We really find they're so enlightening and so very important to bring issues to light. And you're right, we've grown accustomed to documentaries going through rigorous standards and research and fact-checking, and so uh, and it's not been easy to get people to believe them. And now, just recently, this month, there is a Swedish film called Watch the Sky, and that has had the uh language change from Swedish to English. And so this shows you what can happen, uh, meaning that you could take a picture of any person in the educational field, the historical field, entertainment field, and you can take that picture and put it in an AI that will allow it to say anything. So, how we keep up with what's true and not true is a very important thing to work on for the future, right?

SPEAKER_02 7:23

Well, absolutely. I mean, and I think most people at this point would say they don't really trust media that is on the internet, they don't they don't trust what's going on, but I believe that so far we still have trust in documentaries, but that's a very easy thing to be lost. Um, and and trust is an easy thing to lose and a very difficult thing to build to gain back. And when you sit down to watch a documentary, you make certain assumptions, and one of them is that what you're seeing is true. It's truth from a point of view, absolutely, but there is a basic truth to it. And um, you know, it was it was something that was witnessed by a human at a particular moment in time, and that a human-bearing witness is part of that. And when something is created out of AI, that's absolutely not the case, which was something we considered a lot when we were writing the guidelines. But, you know, Gen AI is it's changing our understanding of the future, but it can also change our understanding of the past, and it's really brought with it the ability to rewrite our history, um, both through the creation of fake historical imagery, but also through sowing doubt in authentic historical imagery.

SPEAKER_01 8:35

So my goodness. All right, well, how does the APK define ethical use of AI in documentary filmmaking and what core principles guided this development? Right.

SPEAKER_02 8:51

So I wouldn't say that we specifically define ethical use. I think that might be a little broader. We set out a bunch, we set out a set of standards and a framework for people to use as they consider their usage. I mean, one thing to remember is that the technology is changing extremely rapidly. It keeps evolving much faster than we can keep up with it. So we didn't want to write guidelines that would become obsolete, you know, in a month or six months or something. I mean, we will update them periodically, but we wanted it to be broad enough that people could consider um the basic principles and then look at their own uses within those. So we identify four basic areas for people to consider. The first one is the value of primary sources. Um, and you know, primary sources are, we use that sort of interchangeably with archival evidence, actual archival. Um the second is transparency, and that's really the heart of the guidelines. You never want your audience to be confused about whether what they're seeing is real or synthetic. Um, we know that people are using AI creatively. We don't want to, you know, we don't want to indicate that there's something wrong with that, but there's something wrong if your audience doesn't know that this is a create, some a creative element rather than a historical element. Um, the third thing we talk about is legal issues, which are also evolving very rapidly. And we want filmmakers to stay on top of those and understand what the potential implications are, and that what the implications are when they begin production, if it's a long production, and I've worked on films that take 10 years to go from pre-production to being aired, that those could be changing over the course of the production. So it's important to stay on top of it and work with a lawyer who who does know what's going on. And then the fourth one is the ethical considerations of creating human simulations. And that kind of gets to what you were talking about earlier, Carol. You know, we sometimes call human simulations deep fakes. Um, and what that means is taking a real person, a historical, someone who existed, and having them say or do something that they didn't actually say or do. So it's creating a fake voice, a create, a fake body, a fake face, but you think, you know, the audience might think that they're seeing this actual person. And we felt that even beyond transparency, there's such a great risk in doing that, that even beyond transparency, there are other ethical things that you should be thinking about when you're doing that in the context of a documentary. So those are the four basic ones. And then each one has a lot of issues to consider within it. Um yeah, that's so I don't know if you want me to keep going with that.

SPEAKER_01 11:34

Yes, well, I'd like to understand because you talked about verification, um, the value of the primary sources of so how did the filmmakers tell the audience this is real and this is not, or this we created? How do they do that? How do they do that?

SPEAKER_02 11:52

What is yeah, what are the methods for transparency? Is that what you mean? Yeah. So we're definitely not prescriptive about methods of transparency. And I can't imagine a filmmaker who would even read the guidelines if we were, because you know, filmmakers own their their creativity and their their artistic vision. Um, we give some, but but the bottom line is just that the you don't want the audience to be confused. The so there are a lot of different ways that filmmakers are discovering that they can be transparent. Some of the ones we that we mention in there just to give ideas are it could be integrated into the visual language that you're building for your films. So we've seen filmmakers playing around with using a specific color wash on material that has been generated or altered using generative AI. Um, we've seen people use a special frame around it. So the audience know, you know, accuse the audience that this is a distinctive creative element within the film. It's not historical footage. Um we've also seen use to great success times when filmmakers have integrated the disclosure into the body of the film, into the story itself. Now, some stories lend themselves to that and some don't. Um, one film that I think does this to great success is a film called Another Body, um, which the filmmakers were incredibly intentional and used, you know, really spectacular care in their use of generative AI. So it's a documentary, it's by Sophie Compton and Reuben Hamlin. They're the filmmakers, and the film is about deepfake pornography and the people who have been harmed by it. And they had some of the people who had been harmed by it within the film, but they didn't want to expose their identity. So the filmmakers used the same technology that was used to create the deepfake porn as a way to obscure their identity. Um, and about 13 minutes into the film, they have the characters themselves reveal, they say, you know, what the story we're telling you is completely true, but the face you're seeing is not my face, and the voice you're hearing is not my voice. And it's very powerful for the audience. It in no way detracts from the story, it really deepens the story. Um, you know, obviously, not a lot of films are going to lend themselves to that, but often a narrator can, you know, or or a character within the film can say, because another use that people have is um from there are a lot of groups of marginalized people whose haven't had a lot of archival material created about them, either because it's been suppressed or it never existed in the first place. And in order to tell certain stories, we know that filmmakers want to create Gen AI to re to be able to show some histories that haven't been seen before. You know, and I think acknowledging that within the context of a story is also powerful, you know, for audiences to understand just how much um, how little exists about the certain histories. So I don't think it has to be seen as a detraction from a film. I think it can, the disclosure itself can even be additive. Now, another method, the most common thing that people are doing is either including a just um a card at the beginning saying that generative AI, and we say gen AI sometimes, generative AI was used in the making of this film or to create certain elements, either in the front or the back. And that's great. I think that's that's important, but we rarely think it's enough to just have that. And the reasons are that, first of all, people don't always tune in for the beginning of a film or stay through the end credits. So they may completely miss that piece of information. And then the other problem with it, or the other limitation of it, is just that it's not specific. So it may even confuse the audience more because they know now that generative AI has been used within the film, but they don't know where or for which things. So then they may not trust other pieces that are authentic. So that's that's kind of the limitation of that. And we we do get a little more specific in saying, you know, some uses of generative AI are more concerning than others. So let's say that you use it just because you don't want to pay for stock footage to show a shot of an ocean. Okay. Well, that is generated, it's not real, but it also doesn't necessarily change our understanding of history. And so in a case like that, maybe it is enough just to have something at the front saying some of the B-roll in this was created by using generative AI. But if it's more specifically taking someone from history or showing a historical moment, then we really think that it's important to be clear that that clip, that moment, that photograph, you know, is known, that specific moment is known to be generated. Does that make sense?

SPEAKER_01 16:44

Yes, yes, to pinpoint it, get really specific about it to make people feel comfortable. Yes, I would appreciate that if I was watching something. Uh, and I I think that would be uh uh make you feel that we're back to uh honesty here with the documentary filmmakers because they go to great lengths to give you information for you, mainly you, to make a decision what you think about this issue, right? They show both sides, mostly.

SPEAKER_02 17:17

And I think I mean the last thing any documentary filmmaker wants is to lose the trust of the audience because then you lost them. You put all this work into making a documentary film, and if the audience doesn't trust the person who's walking them through the story, the filmmaker, then you know that completely undermines the film and and the point of the film. And I, you know, I don't think any any no documentary filmmaker wants that. Um, and and we want people to believe that there is something different in a between a documentary and a docudrama. I mean, there are other there are other forms. There's historical fiction, there's docudrama, and those are all totally legitimate forms of filmmaking. They're great forms of filmmaking, but they are different, and people, the audience comes in with different assumptions. And I think when you're making a documentary, you want to speak to the assumptions that the that the audience is coming in with. They sat down specifically to see a story where the filmmaker was being very honest and transparent with the audience about the historicity of what they're showing you.

SPEAKER_01 18:19

Right. Did you want to explain any more of your guidelines specifically?

SPEAKER_02 18:25

Oh, sure. Um, well, going back to the value of primary sources, there are a few things that we want people to think about. Um the first is uh how the generative AI that you're creating might impact our understanding of the past, because synthetic material that is indistinguishable from true archival risks being passed along. You know, moments of documentaries are always, we see this all the time, pulled out and they make their way onto the internet and they make their way into educational materials and into additional, you know, into other documentaries, and they they will then be taken as truth because they've been shown in your film. Your film is part of the historical record. So once they get pulled out and moved into, you know, the general info stream, they they are suddenly truth and they can change, they can forever muddy the historical record. Um, the second thing we advise filmmakers to think about is the role played by algorithmic bias, um, both in reinforcing stereotypes as well as um over uh overcompensating or over-correcting to combat stereotypes. And we've seen that happen with the technology. And each software or each each different type of generative AI, each company has different biases written into it. So the way that generative AI works in a large language model, which a lot of these things like Midjourney and you know DALI and all these are, is that they've scraped, you know, enormous amounts of information from the internet and from you know what the media that's on the internet. Now, one of the things is that as archival producers, we know only a very tiny amount of the archival media that exists is actually on the internet. So that is you're just going to be getting a very small sample that already is going to be biased. We know that the archive, even more generally, in its completion, is also biased. And then, you know, you put that on top of that, just the part that's made it to the internet, and then the way the um whatever the biases are of the actual software. So often stereotypes get amplified. And that's a problem. We've seen that, you know, over time we keep checking back in with the software to see how much bias is in there. Um, and you know, and the companies are aware of this too, and they keep playing with the algorithms to change it. And so some of the things we'd seen, like when the film that I mentioned earlier that we had been working on, um, this was a year, almost two years ago now, and all of the women that were created in it, they looked like uh Angelina Jolie, like that very specifically Angelina Jolie. You know, and there are supermodels and actresses are very disproportionately represented in the depictions of women on the internet. So that was what was being created. Um, so there's that level of problem. There's terrible racial biases that we've seen, you know, coming up. There was a time, yeah. I mean, I won't even repeat them because why do that? But that, you know, we've seen those things happen when you put in a certain prompt that's very general and it brings up a certain race or a certain you know gender for those things, and that's problematic. And then the overcorrection that we've seen happening is, you know, there was this one moment, and it's maybe it's comical, maybe it's not, but like Google ended up having to apologize because uh their tech, their software book Genesis was creating these uh images of African-American Nazis, which is ridiculous because they were overcorrecting to make sure there was representation from all races in everything. So the the point being, and then one thing I'll mention about bias is when we were writing the guidelines, we really struggled with well, humans are biased. We know that, absolutely, and there's biases in all media. So, what's really the difference between human bias and algorithmic bias? And one of the things we came up with that I think holds is that when a human, a human piece of media shows bias, you know who the author is and you can hold them accountable and you can wrestle with their bias and argue it and debate it and contextualize it. But when there's an algorithmic bias showing in a piece of media, you know, a machine-made piece of media, there's no way to do that. There's there's no way to take apart where it came from because it's just a blend of a million different things. So you can't engage with it and interrogate the bias in the same way, which I think adds a layer of danger to it. Um, it does. Anyway, so that and then just one other thing we think of we we mentioned um in terms of primary sources is we really believe. The questions posed by generative AI are on a continuum with many other things that documentary filmmakers have wrestled with forever, including and especially those raised by creating reenactments in documentary. That's been happening for a long time, and there's been a lot of debates within the community about reenactments. But and we think it's on a continuum with that, but that it uh poses a greater risk because generated imagery um requires so little time and expense to produce that it doesn't lend itself to the same kind of rigor. If you're putting in the time and money that's required to, you know, film actors with sets and costumes and pay a cinematographer and lighting, you're going to really do your historical research because you don't want to have to go back and do that again. So we, you know, we want people to bring that level of care and that level of intentionality to the creation of a synthetic image that they would have brought to a reenactment. So that's just one thing we and a final thing we call attention to around the issue of primary sources.

SPEAKER_01 24:16

I would say that is very important because it is expensive. Recreations can be very expensive, but yet they really move the story forward. They're important. Sometimes it's the only way to get that point across. And yes, I would think that that would be one of the first things that would go to AI because you could do it so much cheaper. So your point is that they uh you want them to be very careful and maintain the integrity of what their uh their statement is, right?

SPEAKER_02 24:48

And not to let AI and their process, the integrity of the process. I mean, the the thing about AI, we don't want to see I think there's a concern about AI coming in just to solve problems of money and time. Now, every documentary filmmaker has both those problems. You know, there is never enough money for your film. There is never enough time for your film, but especially money. And then AI just falls from the sky and answers all your problems, but it creates other ones down the line. And that's what we really want people to consider. We don't want it to be seen just as a shortcut. There are a lot of wonderful creative uses of generative AI, and I think that you know, filmmakers will be able to do really great things with it, but we just don't want them to sacrifice that kind of intentionality and care and time and consideration. Um, it shouldn't be less rigorous. It can't, if it's just a shortcut, it's going to really, I think, endanger the form of documentary. But used, you know, with the same intentionality and care as you would other tools, it could be great and elevate it, you know. So we're we're not, we definitely don't frame, we are not anti-AI at all. We're just, you know, pro uh rigor and care and you know, upholding all the things the documentary form has always stood for.

SPEAKER_01 26:10

You're looking at the future and watch could happen, uh, and you're very smart to do that, because if we lose our trust in documentaries, I mean that would be the worst thing possible. Because I always am so happy to watch a documentary because I know these are facts. If we if you get your ENO insurance and you go through all the legal procedures, everything in there you can believe. And that's where we want to keep it.

SPEAKER_02 26:39

It that is my point. And and that is a real concern. It's a it's a concern on many levels. Um, you know, the idea of history being rewritten or people not trusting what they see is scary. Um, you know, I mean, most people bel share a belief that the moon landing happened, that the Holocaust happened, that the Kennedy assassination happened. And when doubts are raised, there's always been documentary proof to keep them in check. But in the very near future, that kind of visual evidence could be rendered meaningless. You know, and the scope of the problem is pretty staggering. In a single recent year, there were 15 billion synthetic images generated. And that was like a year and a half ago. So that that number is growing. But for context, that is the same number as were created by humans during the first 150 years of photography. So the deluge of generated content is just arriving faster than we can process its implications. And that's kind of the concern that led us to create these guidelines and also is now having us um, we're starting to work directly with archives themselves too, about um just sort of safeguarding their collections against the infringement of Gen AI, because we don't want anything sewing doubt in these guardians of our collective history, which is what archives really are.

SPEAKER_01 28:04

Exactly. That's very important. Well, I noticed that you've been endorsed by many major players in the industry, NBC, uh Universal Academy, and over 30 production companies. So, how did you do that?

SPEAKER_02 28:18

Yeah, that we felt that was a really important part of our process. We didn't want it to just be, you know, here we are, these like, you know, archival producers setting out the guidelines. So this was done, there was a very involved process. The first thing we did when we started writing the guidelines was we were in conversation with the International Documentary Association, the IDA, and they invited us to present a draft of the guidelines at their Getting Real conference in um April of 2024. So in the writing of the guidelines, we did talk with just numerous, you know, um, fair use attorneys or, you know, like copyright attorneys, you know, those kind of uh those people, um, ENO insurers, actually. We spoke with ENO insurers, we spoke with people in standards and practices, we spoke with a lot of people who were technologists to really understand AI, and we spoke with film scholars and a lot of filmmakers. So we felt that we had thought through a lot of the issues, but you you just you don't know. So we went, then we went to the um the conference and we had a panel that was attended by, you know, a couple hundred people and we got their feedback. And then we rewrote or, you know, we amended the guidelines based on their feedback, and then we sent it out to hundreds of filmmakers and got more feedback. Um, so that by the time we were finalizing the guidelines, which was around June-ish, we already were in conversation with people really across the industry, and we had had their input on a lot of these issues, which we felt was important for building visibility and also trust in the work that we had done. And then we started began asking for endorsements. So we were in conversation with it was a lot of, you know, and you can see the full list on our website of the people and organizations that have endorsed, but that's an important thing because then they're also the ones who are passing those along. And and the guidelines have influenced a lot. At this point, the NEH has said that they've been, you know, incorporating our guidelines into their thinking. Um, we've been consulting with um Sundance and the IDA around the common the common app for funding. We've spoken with funders, um, PBS standards and practices is utilizing the guidance. So even the ones that even places that haven't endorsed are actually pretty involved in conversation with us. Um a lot of them. So that it's exciting. I mean, we feel like because a lot of different streamers have their and and places have their own guidance, but it's very opaque and it's not all the same. So if you're an independent producer, and like I, as an archival producer, work with sometimes on big, you know, films that are being sponsored by streamers, but often on independent films who don't know where their film is gonna end up. And it's a very confusing territory for them because for an independent producer, you're you make your film and then you go to sell it and you don't know what the standards are gonna be of whoever it is you're selling it to down the road. And it may turn out that you you have to go back and redo some things within your film if you've used generative AI, if you don't know what their standards are. So we're working towards having a more universal type of standard.

SPEAKER_01 31:33

And do you see these standards involving into industry-wide policies?

SPEAKER_02 31:39

I mean, that would, yeah, yes, absolutely. That would be ideal. And we are, it's it's a slow process, um, and it should be a slow process. It's not like we think anyone should just be jumping on this. Um, there's a lot to consider. And, you know, we're built, it's an overused expression, but we're building the plane as we fly it because uh, you know, the technology and legal landscapes are just changing so rapidly, and we're trying to change with it and you know, and be in touch with as many people as possible. I mean, documentary doesn't have a central body that makes decisions for it. So there isn't really anything that everyone's agreed upon in that way. It's it's much less formal than that. So we're we know that that's the you know, the the ecosystem, and we're just trying to get it to be, you know, we're trying to start conversations basically and and create a framework for those conversations to happen within.

SPEAKER_01 32:32

Great. Well, what about international uh documentaries that uh that we have no control over, but they'll be distributed in America? How how are you guys handle that?

SPEAKER_02 32:44

Well, honestly, I feel a little less worried because it's it's almost more the other way around. Like everything's looser in the US, like Canada's ahead of us, and England, uh, the UK is ahead of us. Um, so it's more and and like the EU has standards that we don't have. So it's more like if you're making a film in the United States and you want international distribution, it's more likely that your film won't meet the standards there than the other way around.

SPEAKER_01 33:12

They set their standards early then.

SPEAKER_02 33:14

Yeah, they did. The EU was ahead of us, and we've been in conversation with people, certainly from the UK and Canada, about their thinking because they were ahead of us, absolutely. But we're, you know, I mean, these are conversations that are happening internationally. Our guidelines are currently translated into Spanish, French, Italian, Russian, Japanese, and German, with hopefully more languages coming down the pike. And I will say they were the translations were all done by humans, none of them done by AI. That was important to us.

SPEAKER_01 33:45

So beyond AI, are there any other emerging issues in archival production that the APA is addressing?

SPEAKER_02 33:54

Oh, yeah, yeah. I mean, we are also a membership organization, and so we recently came out with a set of guidelines for working with archival producers. And that felt important because people don't know, even within the documentary field, people don't always understand what archival producers do or what the value of having an archival producer on your team is, and so we wanted to lay that out as well as just give tips for how to get the most out of working with an archival producer, how you get the most value added for your production and what that really is about. Um, and so that's exciting. And we uh, you know, we we did a webinar with the Documentary Producers Alliance and the Alliance for Documentary Editors to kind of get it that guide into their hands, which was exciting. Um, Eugen Branig and Deborah McClatchy were the primary authors on that, and they did that. There was a the IDA covered it in their magazine last month. So that's one thing we're doing, it's getting greater visibility. We're also advocating for um archival producers with awards organizations, you know, trying to get more recognition for archival producers in that way. We recently did a pay survey as well because archival producers are always freelance, you know, we're always uh almost always um working in that way. And it's very hard to know, you know, we're all very siloed. And so we really are trying to create a forum for people to come together and share the issues and understand what's happening in other parts of the country for archival producers and protect our field as well, because you know, Gen AI is dangerous for the archival world too, um, you know, for archival producers and for archives for sure. Absolutely. The archives are it's an existential moment for them.

SPEAKER_01 35:42

Well, how can people get involved with you? Should they go to your website and read all of this information?

SPEAKER_02 35:47

Is that yeah, definitely come to the archival producers um alliance website, which is www.uh archivalproducersalliance.com. Yeah. And we have, yeah, you can sign up to be a subscriber. We send out like a quarterly newsletter about the work that we're doing. Um, there's always new things on the website. Um, you know, we are we're membership funded, basically. We have some grants, which is great. We certainly need more. It's a hard, you know, world out there for everyone right now. But we are doing this work. We actually, one exciting thing that we have coming up is we have been building a toolkit for documentary filmmakers who want to use generative AI in their productions. So it's like a companion to the standards. If you don't want to sit and read, and our guidelines are not that long, by the way. They're like 11 pages, it's not onerous, but even hopefully more accessible is the toolkit that we're coming out with, where one of the exciting things is we're releasing a cue sheet that people can use to track their generative AI uses usage, because that's going to be very important for filmmakers to show to their legal team, to their distributors or streamers. Um, we we want to set, you know, really make it really easy for filmmakers to create these cue sheets like they would for archival materials, like they would for music. Um, this should just be a standard part of production at this point if you're using any kind of generative AI. So the the toolkit has all kinds of things like that. It has uh a workflow chart where you really with all of what you should be thinking about at different stages of pre-production, production, post-production, and distribution if you're using generative AI. Um, and the toolkit is due to come out on April 2nd, fingers crossed. We're currently beta testing it with people in the industry, but that should be that will live on our website and is you know free for any filmmakers to use.

SPEAKER_01 37:46

Wonderful. It's free and it'll be on your website April 2nd on. Correct. Oh my gosh. And that will suggest any AI programs that you have found are good to use?

SPEAKER_02 37:59

That's a good question. That maybe in the next phase of the toolkit, we are, yes, that is something we're looking for doing. We do have resources. We have a whole special resource section that may point to aggregators for the software. We have not done that yet. Um, you know, and and it's complicated because there are so many different issues to consider when you're choosing a software. It's, you know, there's the underlying copyright of the materials that the soft that the Gen AI has scraped from. And sometimes that's known and sometimes that's not known. Large L you know, LLMs don't. There's a lot of exciting work being done with small language models, I think, which will be a little bit cleaner in that way. Um, but we are not at yet a total clearinghouse for all the ins and outs of each different type of software. But that that is part of what we're looking at.

SPEAKER_01 38:54

Yeah. So you've just thrown something uh out that's incredibly interesting. In other words, you really have to research the program you're using to find out what the source of their information is.

SPEAKER_02 39:08

Absolutely. Yeah. I mean, there's a difference between something, you know, something like a mid-journey that's just scraped from the internet and something. I mean, I don't want to, I guess I shouldn't call any specific things out, but there are places that are offering generative AI where they are guaranteeing that everything's been pulled from licensed media or media that they own the license to. So they're kind of guaranteeing you that this was sourced on all, you know, your your copyright is free and clear. And, you know, the the law is not settled yet on how much of a problem it is if you've used Gen AI that is scraping from copyrighted material. There's a lot of cases, and we that we do follow. We have some legal trackers that are tracking the the legal cases as they're coming out. Um that is one of our resources, and that'll be important to follow. But as yet, the law is really unsettled. Um, and you know, and different, and and like, you know, I mentioned that streamers don't necessarily are being a little more opaque. I think they're really watching the laws closely too, because they don't want to find themselves on the wrong side of anything either. Um, but a lot of it is unsettled, and you sort of have to make your decisions before about your what AI you're using before those are settled. So I guess, you know, a lot of people are trying to go for the safe est thing they can find. Um, and again, I'm we just really stress that it's important to talk with a lawyer at, you know, early and late in your use of this. Um, that's not a step that should be bypassed. And we know it's expensive, but there are some great lawyers dealing with this. And there are also some, you know, we point people to like UCLA has a free clinic around this, and so does Cardozo, I think, in New York, um, that can, you know, help filmmakers pro bono.

SPEAKER_01 40:54

Well, my my attorney, Robert Siegel, highly praises you for the work you've done. Oh, that's so nice to hear. Thank you.

SPEAKER_02 41:02

Thank you.

SPEAKER_01 41:03

So that is good when attorneys are saying, well, you have to go listen to what they have to say. It's really important. I think that's great.

SPEAKER_02 41:10

I love to hear that. Thank you. Yeah. No, I mean, and we have been talking to lawyers, like it was important to us that we weren't obviously, you know. And we we got some criticisms of our first draft from lawyers about ways that what we were saying could impact um the legal landscape for uh documentarians in negative ways around fair use and stuff. And we did change obviously everything that was in there that could, because I, you know, we certainly do not want to do anything that could have any negative impacts on any of that, even though this is not a legal document. That's something I should have said at the head. What we have created is not a legal document, it is strictly an ethical one, but even that we were told can have impact.

SPEAKER_01 41:51

So it has a big impact. Thank you so much for the work you've done. This is true dedication to the documentary field. Greatly appreciate it, Rachel. Well, I thank you very much for the information you've given us. I know our filmmakers will love it, and I hope that uh they'll be going in droves to your website and adhere to your policies. And I'm looking so forward to the uh April 2nd release day for the toolkit. Well done.

SPEAKER_02 42:22

Thank you. Thanks so much. I really appreciate it.

SPEAKER_01 42:24

Okay, thank you, Claire, for the show, and we'll be back uh with more information. Thank you, Rachel. Thanks for having me today.

SPEAKER_00 42:34

Be well, everyone.