The following are the outputs of the captioning taken during an IGF intervention. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid, but should not be treated as an authoritative record.
***
>> CHARLOTTE SCADDAN: Good morning, everyone. It's great to be here. I'm Charlotte Scaddan, Senior Adviser on Information Integrity at the United Nations Department of Global Communications, based in New York at the UN Headquarters. Welcome to this session of the IGF Parliamentary Track, organized by the UN Department of Economic and Social Affairs, the Inter‑Parliamentary Union, and our host, Norway. And a warm welcome also to our online audience. We've got lots of people following online.
The Parliamentary Track brings together parliamentarian, private sector, technical experts, and civil society to address the opportunities in our information environment. Today's hybrid session focuses, as you can see, on the societal impact, from risks to information integrity, such as mis‑ and disinformation. We'll hear from our panelists on an overview of the current state of play and emerging trends, the mechanisms by which harmful content is propagated online, and policy responses around the world. Translation, as you just heard, is available in French and Spanish for this session.
A quick note on today's format. I'll invite each panelist to respond to a question from me. We'll have six to seven minutes for each answer, so panelists, bear that in mind. After the panel, we'll have an open discussion. My colleague, Celine, will be helping coordinate the online interventions. And then in closing, I'll invite each panelist to share one concrete takeaway from their final minute.
So, we have an excellent lineup of speakers today. I'm very excited to have them with me. Joining us online, I think ‑‑ I cannot see them ‑‑ but is Lindsay Gorman, Managing Director and Senior Fellow of the Technology Programme at the Transatlantic German Marshall Fund of the U.S. Abdelouahab Yagoubi is Member of the Parliamentary Assembly of the Mediterranean and Rapporteur on Artificial Intelligence of Algeria. Tateishi Toshiaki is from the Japan Internet Providers Assocation. Dominique Hazael‑Massieux is Vice President of Global Impact at W3C. Camille Grenier, Executive Director of the Forum on Information and Democracy. And last but not least, Marjorie Buchser, Senior Consultant, Freedom of Expression and Safety of Journalists, my colleague at UNESCO.
So, to set the discussion going, I'm going to give a few short remarks to provide some context. A year ago tomorrow, in fact, the UN Secretary‑General Antonio Guterres launched the UN Global Principles for Information Integrity. This multistakeholder framework for action took shape amid growing risks to the integrity of our information ecosystem, our global information ecosystem. These risks include: Misinformation, disinformation, hate speech, media suppression, and lack of access to reliable information, all of which undermine human rights. Add emerging technologies, and we can find the pace of these risks now accelerating, their scope expanding ‑‑ thank you ‑‑ and their impact deepening, particularly on vulnerable and marginalized groups and during times of crisis and important societal moments, such as elections.
In this rapidly evolving landscape, the UN Global Principles provide a foundational reference point, and the five principles are: Societal trust and resilience, which involves building resilient communities that can withstand risk to the integrity of the information ecosystem; the principle of healthy incentives focuses on innovating business models and engaging advertisers to demand transparency on where their ads are placed online; and the content online that their ad budgets support.
Public empowerment ensures everyone has the tools and literacy to engage safely and confidently online and can gain better control of their personal data. Independence, free, and pluralistic media supports a diverse range of trustworthy media voices, free from undue influence or censorship. And finally, the principle of transparency and research promotes openness about how digital systems work and supports evidence‑based policies. All of the principles are anchored in a strong commitment to upholding human rights, and of course, freedom of expression.
The principles and their accompanying recommendations call for multistakeholder action. In the last year, since launching, we've seen a hugely encouraging response to these calls to action, with momentum and energy on many fronts. We're bringing a range of actors together, such as governments, civil society, media, academia, the private sector, and local communities, to implement the principles with relevant solutions that meet different information integrity needs.
A key example of this is the Global Initiative for Information Integrity on Climate Change, led by Brazil, the UN, and UNESCO, along with civil society actors, including some with us today. Tomorrow, in the Conference Hall at 3:30 p.m., you can engage with a diverse panel on how to strengthen climate information integrity in the lead‑up to COP30, and I welcome you all to join us. Cami here will be moderating that panel.
The UN information integrity is advancing through the Global Digital Compact agreed by all UN Member States in September. Of particular reference is Action 35E, which focuses on strengthening information integrity to assess, and thereby, support efforts to ensure that the Sustainable Development Goals are not impeded by mis‑ and disinformation.
So, to today's session. It will explore some of the themes that I've just touched on around the risk to information integrity online, including, of course, the impact of emerging AI technologies. So, to get to our questions.
My first is to Lindsay Gorman, who I still cannot see but I believe is with us online. Okay. How do emerging technologies alter the information environment? And what impact do digital technologies have on information technology? What are some concrete recommendations, especially for legislators and policymakers, on how to implement and innovate democracy‑affirming technologies that protect and promote democratic values and human rights? You have the floor.
>> LINDSAY GORMAN: Thanks so much. And I hope you can see me and hear me now.
>> CHARLOTTE SCADDAN: Yes, we can. Yes.
>> LINDSAY GORMAN: Now I can see myself, so fantastic. Yes, thanks for having me, and good to be with you virtually, and what a great list of incisive questions there.
Maybe I would start with very briefly talking about AI's impact on the information environment, as you've asked. I think it's probably not going to be news to too many folks in this room that AI is dramatically transforming the information landscape, I think right before our very eyes, and sometimes without us realizing it. The ability to create and, really, the democratization of this ability to create realistic video, audio, in addition to images and text content that is fully generated by AI has exploded, and it is absolutely impacting democratic processes and communications in democratic environments.
And let me just say, probably also ‑‑ which will not be a surprise to anyone in the room ‑‑ that we're starting from not a great place in terms of a very congested, polluted information environment, even before the addition of artificial intelligence. And so, adding AI as another layer onto that confusion and pollution, where it's very hard to discern fact from fiction is just another additional complication. And we at the German Marshall Fund have done some recent research and analysis, actually tracking the spread of deepfake campaigns around global elections. We took last year, 2024, as a historic election year, and we tracked where deepfakes were happening around the globe, such that they made it into significant reporting in English language. So, obviously, not every deepfake campaign that's happened everywhere, but we looked at these major campaigns that were large enough to attract significant media coverage, and we found that over a third of elections last year had these major deepfake campaigns associated with them, and we found 133 and counting instances of these big deepfake campaigns specifically around global elections.
So, these were things like politicians who had been deepfaked in compromising positions. We had deepfake audio campaigns of politicians' fake claims on a fake recorded audio of tweaking the election result and tampering with the election. In one case, one candidate was faked to have said that they wanted to raise the price of beer, if elected.
On the other end of the spectrum, we saw instances, such as in Argentina, of candidates using AI to create their own campaign posters and campaign messages in kind of an artistic way and paint themselves in a different and interesting light. So, these things are no longer the province of science fiction. They're absolutely in the real world.
And I would just note, the speed with which we went from, we should be worried about deepfakes, oh, but maybe deepfakes are too big of an overhype and they're not actually happening; it's really about cheapfakes and deepfakes aren't really in the real world yet, so how much do we worry about them ‑‑ from that conversation, which was happening just years ago, maybe two years ago, to where they are today, where they are, I think, our research shows, a fact of life in modern‑day elections, has really been 0 to 60 in a nanosecond there. So, these things are everywhere.
And it's not just AI‑generated content, to intend to deceive; we're also seeing the rise of all other kinds of AI platforms that we're starting to communicate with, whether it's AI friends or AI work agents, and these are all going to impact the information that we receive and the type of content that we can access, and ultimately, the trustworthiness, and of course, the sort of business imperative, I would say, for collecting more and more information to train these models and the enabling of a surveillance state that we're very much seeing in countries like China being built out around the world, around the digital silk road. So, these things are only going to accelerate. They're only going to be more essential to modern life, in some ways, as these tools become more useful, and that will have, I think, some dangerous impacts on not just human rights, but also this already congested, polluted information environment.
So, with that sort of downer note to start with, what can we do about it? Part of the work that we've been doing at the German Marshall Fund has been to promote the idea and the innovation and the adoption of these democracy‑affirming technologies, which we did define as technologies that have democratic values built into their very core. So, what are these values? They're things like transparency, privacy, accountability, so that our thesis is that we need to be democratic by design; that the next generation of technologies has to be built with our democratic values at their very core, or they will not support thriving democracies around the world.
And I think we've seen that. We had a naivete that with social media, these technologies would be inherently democratizing because they were inherently connecting people. And we saw very early on in the Arab Spring how protests were gaining steam online, and we thought these would be inherently democratizing forces. And I think we can say this has not come to pass, and probably on balance, the effect of these technologies on democratic values and governance has been a net negative.
I'm not sure this video feed is still working. Hmm... Seems like I've dropped off. Okay, well, maybe I'm back now. Okay, well, I guess I'll just keep going.
So, yes, our thesis is that, ultimately, we need to be building democratic values into technologies from the get‑go, and that's going to take massive societal and entrepreneurial change, because right now, that's not what we're doing. And it's not just about regulating technologies. Okay...
(Zoom technical difficulties)
>> CAMILLE GRENIER: We had a consultative process with more than 400 experts consulted around the world, reaching out to private companies, global papers, and we arrived to three thematic working groups, one on AI, one on media, one on data governance, which is a critical issue, and all through the lens of mis‑ and disinformation.
The report, it is quite long, so I will not give a presentation of the 250 pages, but it covers, indeed, 3,000 academic sources from 84 countries, and twice to answer to almost 40 research questions in nine chapters in 300 pages.
We have different items coming from this Meta report, including summary to policymakers, executive summaries. We have one specific report on future research priorities. We also have mapped ‑‑ all this is on our website ‑‑ we have mapped the research on all of these issues, and we have an interactive bibliography online.
Getting to the conclusions now. First ‑‑ and we should not have to restate this ‑‑ States have a duty to protect human rights and fundamental freedoms. And really, we saw that research consistently emphasizes the need to differentiate between normative goals and principles at a global level and how these are translated into practice at the regional, country, and local level, in ways that fail to uphold the duty of States. This means very concretely that criminalization of the spread of mis‑ and disinformation may not be an option, as it poses direct risks to the right of freedom of expression and really, clearly, other solutions exist.
The second conclusion that we got from the report is the consensus that big tech business models prioritize data monetization for profit. And these business models create dependencies for private and public organizations, as well as individuals and facilitate the weaponization of information, making social media attractive targets for mis‑ and disinformation campaigns that are incompatible with a diverse plural and public sphere.
Third, the exclusion from an inequitable inclusion and information ecosystems at the local, national, and regional level. It is persistent and associated with monopolistic power of big tech companies which leads to harmful discrimination and exclusion.
Four, transparency and accountability. I think it has been said already and will be mentioned again, these are some of the core principles that we have at the Forum on Information and Democracy, and these are measures that are essential to mitigate the harms of mis‑ and disinformation. Research demonstrates the need to enforce big tech company governance to promote AI systems transparency, especially using independent audits and to ensure that accurate information reaches a wide range of stakeholders.
And last, but maybe not least, media and information literacy and AI literacy training is crucial, but it is not a stand‑alone answer to mis‑ and disinformation problem, and we clearly need a more systematic evidence of the outcomes of these initiatives globally and over time. And we've noticed, as well, an insufficient attention to children's literacy.
During all this process ‑‑ so, these are the five main outcomes of the meta‑analysis, but we also addressed the need to strengthen research on mis‑ and disinformation and on information integrity. The first one, first thing that came is to address the western bias. We have seen how research is concentrated in basically Europe and northern America, and we need more research from ‑‑ thank you ‑‑ from the global majority world, and we're really trying to address this.
We need a multi‑dimensional research that addresses the complex component of information ecosystem, something that we've shown is how different from societal to technological to political dimensions, need to be addressed. And last, but not least ‑‑ and we've been advocating that and a lot of people in the room have been advocating for more access to data. So, we really need to build a framework to ensure that researchers, civil society, journalists have access to more data from platforms so that we have a better understanding of what's actually happening in the information ecosystem. Thank you.
>> CHARLOTTE SCADDAN: Thank you so much, Camille. And I think you touched on so many excellent points there. You stressed many things, including how big tech is used to facilitate the weaponization of disinformation behaviors, the importance of transparency and accountability, on which I think we all agree; that media and information literacy is crucial but not a panacea, and I think that is a point that is really key; and of course, your point on strengthening research outside of the English‑speaking western countries I think is something that we're all very concerned about. It was one of the key points in the Global Principles as well, and one of the key objectives of our Global Initiative on Information Integrity, really to kind of get outside of that bubble and find out what's happening in information ecosystems globally.
So, if you haven't read the report, please do. It's an excellent contribution to the information integrity space, and really thanks for all the work on that. So, I believe that Lindsay is back with us. Lindsay, do you hear me?
>> LINDSAY GORMAN: Yes, I do. Do you hear me?
>> CHARLOTTE SCADDAN: Yes! Excellent. I'm sorry that we lost you. Actually, we were still following the script of what you were saying, which was appearing here on a screen, but we couldn't see or hear you. So, I think you left off just when you had actually turned to some of the solutions to some of the AI risks that you heard so eloquently outlined. So, perhaps you just want to in the last couple minutes finish that.
>> LINDSAY GORMAN: Absolutely, yes. Thank you so much. And I will also drop in the chat this report that some of these comments and ideas have been encapsulated in, that we just put out, actually, last month on democracy‑affirming technologies.
And so, our thesis is that we need to be building these technologies in from the ‑‑ these values in from the get‑go to the next generation of technologies. And you know, the unfortunate thing about some of the technologies and the business models that we have today is that they are not very democracy‑affirming. But the good thing is that with technology, every generation of technology is a new opportunity to create something different and to try something else out, and it's never static. So, what we have today is not what we're going to be stuck with for the next 10, 20, 50 years. We always have a new opportunity to innovate, and technology's always racing ahead. So, can we get in at the get‑go, very early on in the technology development process and build the next generation of technologies that go viral, to be explicitly in support of democratic values.
And this is not to say that regulation doesn't have a role, that policy and governance, all of these things have roles in creating a better democratic technological ecosystem that supports our values and our security, but that innovation also should play a role, too. And so, that's what we outline in this report, where we've conducted two pilot projects on innovating and adopting democracy‑affirming technologies. The first on innovating was, we actually held a Hackathon for teams of entrepreneurs, technologists; we had coders on the teams, as well as academics and civil society activists. We did this in Mexico City, around the context of Mexico's election, to try to create new innovative technologies.
And we got some very interesting prototypes, and I think the next step really is to build a community of not just entrepreneurs, but also potential funders and investors and policy organs that want to take some of this work forward, not just at the Hackathon that we held, but at efforts that are springing up around the world to do similar.
For example, just this past month, I was in Germany, in Leipzig on the jury for a competition for AI to fight mis‑ and disinformation, deepfake detection and prevention. And there are some incredible technologists and ideas and solutions that are popping up. None of them, I don't think anyone thinks they're going to be a panacea, but innovation needs to be in this game as well. So, there are a bunch of efforts here. And what we really need on the policy side is significant oomph we need from funders, from philanthropic organizations, from investors, investments in these technologies.
One of the recommendations in our report is also to create research scholarships around for young researchers and entrepreneurs who want to build these technologies. Is there a career path around democracy‑affirming technologies, other than just joining one of the large technology companies? Can there be an innovation in the public interest career effort? And that's where I think a lot of governments and philanthropic organizations and advocacy organizations can really come in and create a career path for younger folks starting out in the technology field who do want to build democratic values in.
The second project we conducted was one on adoption of democracy‑affirming technologies. And in this one, also in the context of Mexico's election, we partnered with two media organizations ‑‑ one photojournalist agency in Mexico City, and a larger international media organization, the Canadian Broadcasting Corporation to build a repository of images around Mexico's election.
So, what we did is we worked with Obturador to take these amazing photos. Really, they took all the amazing photos of Mexico's election last year, including some fantastic images of now President Claudia Sheinbaum at the polling places, at the boxes casting her vote. And then, we partnered with technology providers ‑‑ Microsoft and Truepic, a smaller company working on content authenticity and Providence technology ‑‑ to authenticate the images and create a tamper‑resistant record of how these images have been taken; whether they've been modified along the way, from the camera to your social media feed. And we've posted this online, and I'll drop the link in the chat as well, as I think the first sort of tamper‑resistant repository of election imagery.
And also, got one of these photos featured in the Canadian Broadcasting Corporation's coverage, actually, that they took of Mexico's election with a content authenticated image logo and all the metadata that goes with that, actually, in CBC's coverage. So, we're moving slowly, I think, on the adoption front, but these sorts of pilot projects can help us take this next step towards technologies that better promote democracy and democratic values.
And for example, with content‑authenticated images, these really promote transparency so that, essentially, you're getting kind of a digital ingredient list of how an image has been created, how it's been modified, how it's come to your feed, so that as the user, you can decide for yourself whether to trust it. I like to think of these things like text. If we read a text that's been very well cited, where all the assertions made in the article can be very well referenced and traced back to the original sources, and we can actually go into those sources and see, does the claim made in the original article ‑‑ is it actually supported by what's cited? That builds trust, that sort of transparency builds trusted information.
Whereas, if we read an article with no citations, with claims that come out of nowhere, we don't really know whether to trust it. And so, this is what we're trying to do in the visual information environment with video and with images on content authenticity technologies. So, this was our pilot project. I'll drop the link in the chat so you can see some of the images that are authenticated. But this was in our sort of realm of adoption.
And then, just quickly on some of the policy recommendations, both in the innovation and the adoption, we recommend that online platforms and websites should be incorporating some of the existing democracy‑affirming tools, like content authenticity technology into new products and new product design; other technologies in our democracy‑affirming tech suite include privacy‑enhancing technologies, some of the censorship circumvention technologies that are used around the world to counter government censorship. Government agencies, private sector champions, philanthropic organizations and universities should create these democracy‑affirming tech research scholarships, just for innovation for a new generation of technologies.
Governments also should be working with technology providers and even the hacker community to red‑team these emerging democracy‑affirming technologies to shore up any vulnerabilities and move them from concept to true adoption, then also should provide guidance on how to use these technologies and how these implementations can advance cyber and national security, in addition to democratic values.
So, there's a lot to be done, but I think that, yes, every new generation is a new opportunity. There is definitely steam around the movement for content authenticity technologies and, you know, can we as a global community give some of these technologies an oomph, as well as create the space and the incentive, really, for younger folks coming into the technology field to make their career and life's work on building the next generation of these technologies to support and enhance values that we would like to see in our information environment. So, I'll stop there, but thanks. Thanks for the discussion and looking forward to the next speakers and questions.
>> CHARLOTTE SCADDAN: Thank you so much, Lindsay. I'm glad we managed to get you back, because it's really important to focus on solutions and not just the risks. And I really appreciate your focus there on the innovation that's going to be needed in this space and how we are going to have to attract the talent that we need to take those effective solutions forward.
So, without further ado ‑‑ and I just want to let everyone know that we are able, because we started a little bit late and we had a few technical hiccups, to extend the end of the session. So, we can go to 11:15, so rest assured, there will be plenty of time for questions. I don't know if all of our panelists can stay, but they don't have much choice because I've just said it. And otherwise, they're kind of trapped here, so, hoping they can.
All right. So, moving on to Marjorie. UNESCO is the UN agency in charge of the promoting of the safety of journalists. So, what are some of the trends to counter risk to information integrity that you're seeing? What are some of the impacts that you're seeing in the space on freedom of expression on the press and on access to information, and what's UNESCO's role in the response to that?
>> MARJORIE BUCHSER: Thank you very much, Charlotte. And it's a pleasure to be back at the IGF in Norway, which is my first time here and really looking forward to visiting a little bit more afterwards.
First, I need to preface, if you have the impression that I'm catching my breath and I deliver my remark in a slightly slow fashion, this is true. I'm currently five months pretty good, so I will take my time because I am speaking for two. And Charlotte, do feel free to, you know, let me know when we are time and I'll make sure that I'll wrap up.
So, as Charlotte mentioned, UNESCO is a specialized UN agency, which holds a mandate of promoting and protecting free flow of ideas, which includes freedom of expression, access to information, and safety of journalists. And of course, digital technologies have really kept us busy, because this is really a changing space.
Now, there are many trends that I could have focused on, but I will, like Lindsay, look specifically at advanced AI models and generative AI, which is this category of AI that sort of generates seemingly new information on the basis of a simple question prompt from a user.
And what is, I think, very important to realize in this space is that, while you may have a lot of headlines about generative AI, this is very much a technology in a nascent stage. And if you read the scientific report on generative AI, its impact and how to manage it, I think there are two main points that are important to mention. The first one is that it's changing and performing drastically differently, almost on a monthly basis. So, you have this rapid evolution, coupled with deep uncertainty on how actually to manage and mitigate it. So, if you have the impression that you can't follow, you're not the only one. Even experts, technical experts in this field, struggle with the pace and the lack of real evidence on this technology.
However ‑‑ and I think that's why we're here ‑‑ there's a strong consensus that has emerged that this technology has a particular impact on information integrity. And if you look at the UN last report that was made by the high‑level advisory body on AI, most of the experts ‑‑ 80% of them sort of agree ‑‑ and that's across background and across region ‑‑ that information integrity is likely to be highly damaged by, or has a strong risk with advanced AI models.
So, what I want to do is talk to you about three specific challenges and few governance solutions and specific recommendations that UNESCO has put forward. This is also the topic of one of our latest reports on generative AI.
So, the first one ‑‑ and this is something Lindsay and as well as Camille mentioned ‑‑ is that obviously, AI increases the risk of information manipulation. Lindsay mentioned all the deepfake campaign, but it's also the convergence with generative AI, where it allows you to create fake content very easily, and digital platform that allows you to disseminate it to a really broad audience. That is the key problem. And so, what you have is increasingly confusion online on the authenticity and the officialship of the content that you see.
I also want to invite you to think about it not as a dichotomy ‑‑ is this AI generated is not the same thing as is it trustworthy. If you, for example, want to post a LinkedIn comment to say you're in IGF, and you ask one of the popular generative applications to generate it for you, is it not trustworthy? It may be completely true, but it's completely AI‑generated on your prompt. So, the real questions is also about what is the context in which this content has been used? What's the purpose of it? And so, it's really a deeper contextural discussion that needs to take place.
The second challenge is AI has transformed the way we see information. What we see in the younger generation is they're increasingly bypassed established news website to only use recommended feeds or, basically, generative AI application, and this is a major difference, because if you look at traditional search engine, what they will do is be an inter to lead you to traditional or journalistic sources. What generative AI does is it aggregates a different version of this topic and brings it back to you, but it's notoriously bad in citing, in quoting, or references. So, basically, what it does, it removes traffic from established journalistic sources.
And this is also a key problem, because what we see is, increasingly, users over‑rely on AI, so they trust AI outputs even more than some journalistic content. And so, this is ‑‑ you know, we know that AI confabulates, we know that it's biased. Nonetheless, there is a tendency of users not to use it critically at all. And that's not only the average user. What we see increasingly is that people are quoting full spoke or full references because it was generated by AI and not question.
And I think the last challenge I want to highlight before I move on to the solutions, that is, I think for us, potentially the most fundamental one is that AI‑powered application poses significant threat to pluralism of voices and diversity of content at a fundamental level. The AI model that you see today are trained on English‑speaking data from the Internet that is mainly generated in the North. And what it means is that it really sort of aggregates a specific vision of the world with its inherent biases. And so, that is for us one of the most significant threats to the plurality and diversity of the ecosystem, information ecosystem that you have online.
Now, there are many challenges, and I think we'll talk about it a lot, but I think it's really, also, as I said in the introduction, important to remember that we still are uncertain about the trajectory. This is still something that we can shape. And obviously, what UNESCO stands for, it's sort of evidence‑based inclusive governance model. And so, there are three points on governance and principle that I want to highlight, as well as maybe more specific recommendations.
At the governance level, I think there are three really key principles in a lot of our guidelines. The first is the notion of outcome‑based approach. For a long time, what we see is a tendency to try to regulate every piece of content, every algorithm that exists, which is impossible but also has a very chilling effect on freedom of expression. And so, one of the recommendations is really to think about the system. What are the processes and mechanisms that could be put in place that mitigate the most negative outcomes? So, that's one.
The second one, and it's not going to come as a surprise, but it's the notion of multistakeholder approach, but also multistakeholder that is contextual. Today, AI's developed by commercial labs, and it's not necessarily ill intended, but it's a very specific vision of the world with, you know, a very specific culture, and it's very important to have multistakeholder perspective, but also from different local and regional level to input diversity in every stage of the AI lifecycle, from the data to the output.
And the last one, in terms of, specifically on the governance level ‑‑ and I think that's also something that Lindsay mentioned ‑‑ is that multi‑layer approach. And I can imagine as parliamentarians who think of statutory regulation and statutory framework as the initial approach, which is, of course, a very important one. But what we see is that different sort of layer, from co‑regulatory to voluntary commitment can also have positive consequence, and some of them can be very technical, so less prone to regulatory framework and more for voluntary commitment. So, that's at the principal level.
And to finish, maybe three more specific recommendations for generative AI. The first one is really the importance of enhanced transparency. Today, we do not know who uses generative AI, for what purpose, and what's the impact. If you look at the Transparency index that scientific organizations have done, we just don't know. And not only do we not know, but there is also no indicator in terms of how it impacts different regions, different cultures, with really different effect. So, that's a very important element with maybe preferential access to researcher and journalists to the data and the process of tech companies.
The second one ‑‑ and it's something that Camille mentioned ‑‑ is the importance of literacy. As I said, the way users sort of access information is changing. And what is really essential is to help them critically assess the use of those tools and just really trying to understand the limitation of it, and that can come with clear labelling in terms of the provenance, the context, but really trying to help those new users really sort of understand the beneficial use, but also limited, or limitation of the technology they employ.
And finally, I think the notion of public investment and open solutions to really support freedom of expression in all its forms is necessary. What we know is commercial dynamic is insufficient to provide the diversity that we need, and so, it's an investment in low resources in language, cultures, so that it could be digitalized, it could be brought about and there could be data and models based on that. It's also about representation and access, and in its context, open solution that allows more parties to sort of develop maybe more context‑specific models, are very beneficial for those problems. So, that's the different aspect I wanted to leave with you. I hope I didn't take too long, I think?
>> CHARLOTTE SCADDAN: No, no. Thank you. Thank you, Marjorie, for that important intervention and for highlighting those recommendations. And I'm sure there will be questions related to the AI space, because you know, it is obviously a topic that many here are very concerned with.
I do want to make sure we have time for questions on that note, so I will move to our next speaker, Dominique. Can you let us know more about W3C's work? What role can web standards play in mitigating the impact of mis‑ and disinformation and other information integrity risks on society? And what is W3C's roadmap in this space?
>> DOMINIQUE HAZAEL‑MASSIEUX: Thank you. And thank you for having me on this panel. So, let me first introduce quickly W3C, which stands for the World Wide Web Consortium. We are a nonprofit international organization dedicated to building standards for the Web. Basically, we convene a big part of the tech community with the mission of making the Web work for everyone.
Of course, the Web has become this platform that has become key to so many parts of our digital societies, making the Web work for everyone is not an easy job. It's also something that is very fast evolving. We have already had a number of discussions around AI, which is directly feeding from the Web, and also creating new challenges for the Web, which is one of the topics I am in with W3C.
But when you think of making the web work for everyone, W3C has been doing work in making the Web accessible to people with disabilities for more than 20 years, making sure the Web can work across cultures, which is key in making sure that content standards is used within these AI systems, also it is exposed to different cultures, different languages, also different abilities, not just people that can walk, speak, see, but also people that have limitations in their abilities to interact with digital content.
But beyond that, a big part of our belief that the Web is a platform for good, a platform that benefits society, when we want it to work for everyone, mis‑ and disinformation is a direct harm to societies. Instead of having the Web work for everyone, it starts working against people, and that's something, of course, that we don't want to see happen, we don't want to see remain.
Our work is anchored in building these technical standards. So, if you open any website, you would be interacting with those standards. As those of you participating remotely, you're probably using one of the technologies that enables remote communication on the web.
The question when we develop standards for the Web is, what is it that we want to bring scale to? What is it that the platform, the Web as a platform, benefits from, interoperability that we make sure that the solution that we develop is not something that will just work in a specific market, a specific region, or a specific type of environment, but something that will work on a global scale. That's a scary responsibility, but it's one that we take with a lot of passion and attention, in particular a lot of the ways these technical standards are developed, I think are very similar to your experience as parliamentarians, we take a very deliberative approach, analysis of the problem space, analysis of the solution space.
And so, very recently, we've restarted some of our work around this problem of mis‑ and disinformation, trying to understand what recent changes, not only in the ecosystem ‑‑ obviously, the rise of generative AI as a multiplication of content that can be developed through these mechanisms ‑‑ also the new social media ecosystem, which is I think a lot more fragmented than it was a few years ago; new technologies that have emerged. Lindsay earlier related to some technologies emerging in content authenticity certification. So, we are basically looking at this new ecosystem, looking at the solutions that have emerged in that space, so C2PA is one of them in terms of facilitating content authenticity.
We're looking at a research effort from MIT called TrustNet, which is about how using social trust relationship as a way of helping people assess whether they can or cannot trust a particular piece of information, so building on the existing mechanisms that we as humans have used for thousands or millions of years to make oral opinion, to make sense of the world. I assume very few of you actually know how content mechanics works, yet we are using these devices that only work because we all actually benefit from that. So, it's not because we know the theory behind the mechanism that we trust it; we trust it because people trust, have been able to actually make use of these technologies. So, TrustNet is this approach in building up trust in relationship between people to, again, reflect the way humans actually build their world knowledge.
Another technology we are looking at is something from Japan is called Original Profile, again, trying to build on existing social systems that have been built to represent trust into the ecosystem. And I won't go through the full list of technologies we're looking at, but what we're trying to do is take a very systematic approach at reviewing what these technologies enable, understanding the risks that they might bring ‑‑ clearly, we want to fight misinformation; we don't want to create censorship, which is the other side of the coin. If you try to control too much the information ecosystem, you risk depriving it from important diversity of perspectives and opinions.
So, again, we try to take this systematic approach and we devise technologies, understanding which part of these technologies deserves to gain scale, gain end user visibility. Again, one of the roles is basically to bring technologies to the end users who have a sense of understanding which of these technologies can be meaningful to end users, can have a meaningful impact on the way end users perceive the information ecosystem. That's what we are currently doing.
Where we will land, we don't know yet. You probably know that deliberations leads to the committee where it can land. One thing that is quite clear is wherever we land, the standards we develop only make sense if and when they get adopted. We are developing voluntary standards, so it's not like we can say, this is the new standard and everyone will adopt them. They only get adopted if the right incentives exist for the people in the ecosystem to adopt them. And of course, we have a lot of the right people participating in these discussions ‑‑ all the major technology providers are part and active in W3C.
We have lots of content and media publishers. We have lots of civil society, NGOs, government agencies, so lots of people are already involved. But at the end of the day, if we come up with these new technical standards, they will only be adopted if the right incentives exist, including in the market. And that's where I see a critical need for stronger cooperation between the regulators, the policymakers, the legislators, including the parliaments around the world, in making sure that as you develop these new policies, as you develop these new approaches, you're managing mis‑ and disinformation, they integrate, they build on top of those technologies. So, there is analysis we're trying to conduct to make sure that they work together, rather than against one another.
So, I guess if there is one message I am trying to communicate today, that there is a lot of syncing and work happening inside the technical community, and in particular, in W3C, around these questions, and making sure as you develop your own policy agenda in that space, that you talk to us, that you talk with us, that you maybe even contribute to our work, to make sure that at the end of the day, our work complements one another and builds on one another. I think this is the most likely path to success in what is a very complex and multifaceted programme space.
>> CHARLOTTE SCADDAN: Thank you so much for outlining some of the very concrete solutions that you're working on, and indeed, stressing the need for global solutions to what are global challenges and highlighting the importance again. And I think this is a common thread of establishing, bolstering, and ensuring trust in the information ecosystem, and for stressing the importance of market incentives. I think that is part of the reason you mentioned at the end partnership, and I think this multistakeholder approach is really key in this space. So, thank you.
Okay, we have two more speakers left and we definitely want to get to questions. So, moving on to Toshiaki Tateishi. Thank you for being with us. To counter information manipulation, could you share with us the Web Credibility Assessment Challenge that you are now attempting to begin in Japan?
>> TATEISHI TOSHIAKI: Yes, thank you very much. Thank you for inviting me here. So, I'm very glad to. It's my honour.
So, I think that the disinformation and misinformation, content measure, the transparency is the most important thing. So then, how we provide the trusted Internet to the users today, Internet, all the users. So, ISPs tried to make a effort, mostly about 20 years or more.
So, then, talking about the Internet credibility, so we need to make some audit association something, organization should be. So, we have so many technical management issue holder, trusted sometimes, photo or something, other digital content. So, two years ago ‑‑ sorry, 1 1/2 years ago, we had a big earthquake in the Noto area in Japan. So, at that time, I feared something terrible, because someone had an emergency call, but so then, a firefighter tried to get there, but there is no one, so something like that happened at that time.
So, we have to ‑‑ I think that it has many, or a long efforts, but we have to have some trusted, how can I say, measurement and so on. But as you know, the level of users cannot have that, it's not easy thing. So, in spite of that, probably we can do something, make trusted organization, something, to make sure, for example, to go there and we meet someone who tried to, the people will lead the good way. So, I think it is a very, very hard way and takes so much time, but I think that it's a very important thing.
So, in Japan, so sometime, maybe 15 years ago, we have hard to proctor child pornography. So, we tried to resist that system install our network, but we gave up because of the human rights of the children. So as a national people, we do it for that. So, even it is illegal. But nowadays, maybe seven years ago, we have a parent site programme. So, this year, we have an online programme. So many people easy to say that, blocked these sites, or tried to block. But in the mechanisms in the Internet, block something is ‑‑ every time violate your rights about the particular telecommunication.
And in Japan, I think the secret of the telecommunication is the most big pillar or to maintain the democracy in Japan. So, if we take it much more time, we do it, make some eligible and trusted organization. Thank you very much.
>> CHARLOTTE SCADDAN: Thank you so much for highlighting that very interesting and important initiative at the international level.
And last but not least, my final speaker today is Abdelouahab Yagoubi, who will be speaking in French. So, for those of you who need to put on headsets, do so now. And my question to you is, from the perspective of a parliamentarian, what can parliaments do to strengthen the resilience of societies and democratic systems in the context of the misuse of AI and other emerging tech? The floor is yours.
>> ABDELOUAHAB YAGOUBI: Thank you, Charlotte. Excellencies, distinguished colleagues, ladies and gentlemen, it's really a pleasure to be with you and participate to this important panel. And as there is not Arabic translation, and I am not perfectly fluent in English, I invite you to switch to French channel, because it's more comfortable for me. Thank you.
It is a pleasure to take this Parliamentary Track of the IGF on Internet governance. Charlotte put me last. Therefore, I will take revenge and tell you all of my life and take my time.
I am talking as a Member of the Parliamentary Assembly of the Mediterranean, 35 national parliaments on the southern and northern side of the Mediterranean and the Gulf States. So, the topic of this session is absolutely topical for us, seeing what is happening in the region, and, of course, internationally.
We see proliferation of disinformation campaigns and misinformation online with the massive broadcasting of fake content, which are harming our societies, and this is also harming social cohesion, the trust between citizens, trust in public institutions, economic actors, and we have a very harmful general climate in our societies.
The members of the Parliamentary Assembly of the Mediterranean, we have the Euro, Med, and Gulf countries are very much worried. Our Assembly has taken action. We are following very closely these phenomena. We share the good practices and information from our relevant countries. We exchange a lot of data, and we take concrete measures.
Among the tools, I would like to give you a highlight. We are finishing a report called "Resilience of the Democratic System in the Face of the AI Uses in Information Technology, as well as Emerging Technologies." This report has been assessed by experts from research centres and universities in the world, and it is warning us about the external and internal pressures on the stability of our democracies. We must say loud and clear that the manipulation of information is definitely feeding the fears and confusion in people's minds. The more vulnerable among the population are paying the highest price ‑‑ women, migrants, and minorities.
What is even more worrying at the global level, we have systematic and coordinated campaigns of disinformation organized by state and non‑state actors, namely, the Russian Federation, China, and North Korea. It is aimed directly at the democratic processes. It is also compromising the choices of the electors. It is also threatening the results of electoral campaigns. We could see such attempts in France, Moldova, and elsewhere. Of course, it's absolutely crucial to react.
We must first have robust and ethical legislative frames, in order to fight back against these digital contents. We must reinforce our alliances with governments, civil societies, community, and religious actors, and of course, the private sector, following the Universal Principles of the UN for the integrity of information and the Pact for the Future.
We must also, thirdly, invest in literacy. Digital literacy is the best weapon against manipulation and disinformation.
Before I conclude, I would like to put forward two initiatives of our Assembly.
(No English translation)
>> CHARLOTTE SCADDAN: Thank you so much. And you were, in fact, incredibly concise and right to time, and as I said, last but absolutely not least. You had the final word.
So, we do have time now for questions. I believe colleagues will be coming around with a mic, for those of you who want to ask. So, if you could raise your hand so that I can see you. I have already in the front here, and then a gentleman over there and then a couple of others at the back here. So, maybe we start at the front, right here. Thank you. And if you could let us know if your question is directed to a specific panelist, please, that would be very helpful.
The mic is not working. Oh, wait, now I believe I... yep. No. Here we go. We've got you another one.
>> ANUSHA RAHMAN: Yes. Thank you very much. I'm Anusha Rahman, Senator from Pakistan. I have been a former Minister for Information Technology and Telecommunication, and I am now part of the Standing Committee on Information Technology. I am into telecommunications and looking after this for the last 32 years and have worked very closely with Telenor.
So, my question, basically, about the presentations today has a very ‑‑ has an angle which everybody wants to know. When the deepfake is easily possible to be spread on the platforms, my question, frankly, is, why is it so easy for the platforms to make the deepfakes to become accessible on platforms to the point where it is spread like a viral drug? And why is it that the platforms do not wish to cooperate with the countries to move on the complaints that such‑and‑such content is deepfake and should not be there?
So, what we are now fighting with is not so much as to what is deepfake; my observation is that there is deep commercial interest that leads that deepfake to become a account of choice on the social media platforms. Now, to say that the parliamentarians are not thinking of regulation and should think about educating the citizens is not an easy task, when the brainwashing is happening by the social media platforms.
So, my question is, is that this debate which was very insightful, is what are the solutions we would like to see happen for the social media platforms to bring out more cooperation to bring in more stability and return democratic governance to the countries? Thank you.
>> CHARLOTTE SCADDAN: Thank you. Who among us would like to take that question? And, actually, also, Lindsay... I don't know if Lindsay is still online, but I unfortunately can't see her. So, Lindsay, if you would like to jump in for any of these, please do just speak up. Why don't we answer that one first because I'm not sure how much time we have? Go ahead, Camille.
>> CAMILLE GRENIER: I'll try to be brief with a very important question. I feel that we try to work with platforms. They last year, during the election year, they had this tech accord on AI interface. When we reached out to them earlier this year, they say, well, actually, we've seen that we don't have a problem with AI and deepfakes ‑‑ we don't have the AI apocalypse and deepfake apocalypse that everyone was talking about.
So, there is this trend from these companies that say, well, actually, it's not really a problem because we don't have so much deepfakes. That's the first part of the answer. The second part, regarding the economic interest, I think it's an economic disinterest. When we see that if this is not a very important market for companies, they will basically not answer to any queries. And you know, I remember a former head of the regulator in Tunisia, for example, they would tell us that, you know, in Europe, they answer the phones and they sometimes comply, but in Tunisia, they don't even pick up the phone. They don't care. Or they send you to someone who is in a totally different region.
Calling for more accountability, calling for the presence of, for example, moderators in different regions, in speaking different languages, is one of our top priorities, and we clearly see that we have some complete compliance with tech in that regard. It may not fully answer your question, which is a very important one, but I think there are some structural aspects there that need to be addressed as well.
>> CHARLOTTE SCADDAN: Thanks, Camille. And I think what you just outlined there I think reflects the experience and relationship that many of us have now with the big tech platforms, which is that things are perhaps moving not in the direction that we had hoped, especially over the past six months or so.
We do have a lot of questions. Yeah.
Yeah. And addressing those commercial interests I think is key, and that's why I think many of you are here, right? Okay, next question. I think there was a gentleman there in the second row who had his hand up.
>> AUDIENCE: Thank you very much, and good morning, all. First, it's to thank all the panelists today for a great and insightful presentations that have been done. And sitting here as a parliamentarian, the question that is ringing to my mind is that we already are too far out in the development of AI, which is also leading to disinformation and misinformation and wondering, where is a place for forums like IGF to first start out by having a conversation on even what language this technology is speaking. Because as it seems, the languages that are speaking ‑‑ the language models are depending on dominant languages for the development of these technologies, and in that way, we're really fighting against the diversity. And in such a situation, we are asking ourselves, even as legislators from Africa, should we be starting by thinking about literacy of the masses, or should we be thinking about the controls on the development and the deployment in our jurisdictions?
And in that sense, I'm asking the panelists, would they be recommending for countries to take approaches that look at their laws for either repurposing or retiring or developing new laws that are locally developed, rather than depending on the central top‑down presentation as we saw in data protection? We saw the models that we're taking was an American model developing on one side of the globe, the GDPR developing in Europe, and the rest of the world being told to follow in forming their laws in that way.
So, I think Abdelouahab Yagoubi can answer by telling us, are you seeing examples of countries in the southern hemisphere that are making good steps in legislation and regulation that is homegrown and that is speaking our language, that now starts us to feeding the information that goes to power these language models so that then we are not disenfranchised by speaking a language that is not our own, trying to play catch‑up, and borrowing laws that are not our own? Is there a place for me to say, Kenya first, before I think about anybody else, or should I jump onto this IGF bandwagon and say, let's see what is happening elsewhere and copy/paste that? I thank you.
>> CHARLOTTE SCADDAN: Thank you. Go ahead.
>> ABDELOUAHAB YAGOUBI: Yeah, thank you, dear colleague. I'll try to answer in English. Your question is really important. I think in Africa, we have to work together and cooperate to regulate artificial intelligence. Some examples. In Algeria, my country, we have after the European Act on Artificial Intelligence, introduce a proposal to now regulate usage of artificial intelligence in Algeria.
I think Saudi Arabia, with their ITIC platform is a very good model to follow, but I think that we have to introduce a framework of cooperation between African countries and follow the evolution of regulation of this technology all over the world. Thank you.
>> CHARLOTTE SCADDAN: Yeah, thank you. And Camille, you wanted to add.
>> CAMILLE GRENIER: Very quickly, in terms of Global South, I feel that a lot of interesting things are happening. If you look at Brazil, they have ongoing discussions on AI regulation right now at the Parliament. Uruguay is starting some work. Dominican Republic as well. I think, really, Latin America, there is a movement of processes, rather than the content. It is a value interesting development there.
>> CHARLOTTE SCADDAN: Thank you. Okay, I think we might have time for one more question. And I just am aware that I would like some gender equality. Yes. And maybe even some diversity in age. There is a young lady at the back there, if we could please get her the mic. Thank you.
>> AUDIENCE: Thank you very much. I promise I'll keep it short. I'm a member of the Austrian Parliament. I'm wondering also about all of these issues. And in the EU, we do have the DSA and other regulations, so, how do you view that? Does it go too far, not far enough? Is there something missing that we can help with this destruction ‑‑ against this destruction of democracy?
>> CHARLOTTE SCADDAN: Panelist? Anyone want to respond to the DSA?
>> CAMILLE GRENIER: I can be very brief.
>> CHARLOTTE SCADDAN: Go ahead, Camille. Very brief.
>> CAMILLE GRENIER: Sorry. But for parliaments to act, act fast, and really work on the implementation. I was recently in Portugal. They still don't have a law that implements the DSA in national regulation, and I think parliamentarians really have an important role in using this tool that is unique around the world, using the DSA to ensure that we protect our democracies and that we, again, have this structural approach.
>> MARJORIE BUCHSER: Maybe a few words. In terms of the DSA, it has been created in a very specific context, where most European countries have independent regulatory authorities and a set of infrastructure that also supports this regulatory framework. That said, I think from the government principle that highlighted, I think the notion of system‑based or outcome‑based, that looks at, let's look at the mechanism, not at the content, I think it is very interesting.
DSA, the notion of asymmetrical requirement to tackle the issues that the lady mentioned, in terms of trying to identify those big actors that are really structuring the space. I think of an interesting principle that could be, you know, interested, not only in the same system as we see in European state.
>> CHARLOTTE SCADDAN: Thanks. I think important points there ‑‑ legislation is key. Having the legislation in place is an important first step but also has to be implemented. But as we heard, DSA, we cannot take a cookie cutter approach. We absolutely need to look at national context in rolling out.
So, I have, maybe, if the panelists -- I could give you 30 seconds? I think we're a minute over. Just to wrap up with any final thoughts and maybe we'll start at that end and work our way down. Please.
>> TATEISHI TOSHIAKI: So, I think that the legislation is very different from countries, so I think we have to work together. That is the best way to solve the problems, I think. Thank you.
>> CHARLOTTE SCADDAN: Thanks.
>> CAMILLE GRENIER: I'll be very brief, but a good solution to fighting disinformation is also to have free, independent, reliable information and entering that in this discussion we also address press freedom, media sustainability, the safety of journalists around the world, that, you know, we also don't forget those who are fighting on the sidelines and make sure they can do their work freely. Thank you.
>> CHARLOTTE SCADDAN: Thank you and very important point. Please.
>> Abdelouahab Yagoubi: Thank you, Charlotte. I think we have together to profit by the work of the Council of Europe about democracy and artificial intelligence, and I invite you to visit our website, PAM, to follow the digital daily publishing about artificial intelligence. Thank you.
>> MARJORIE BUCHSER: Yeah, I think that, you know, in a context where there's a tendency to say we have to regulate now, I think it's important to think of information integrity, not only about the risk on manipulation, et cetera, but also as an investment. It's an investment, in creating and digitalizing laws, culture and language into creating a local solution, so it's not only cultural, it's also investment feeding in the cultural linguistic diversity to be part of the online ecosystem.
>> DOMINIQUE HAZAEL‑MASSIEUX: Yeah, and I'll reiterate my call for working with the technical community as you develop new regulation in that space, making sure you understand what we are also trying to build in defending the web, the Internet, from this flood of mis‑ and disinformation, and I'll be more than happy to hear from you after this event to see how that could work. Thank you very much.
>> CHARLOTTE SCADDAN: And I don't know if we have Lindsay online still? Lindsay? No? Okay. Well, with that, we'll wrap up. Thank you so much for bearing with us through the technical challenges. We really appreciate it. It was an excellent discussion. Thank you all.
(Applause)
And I believe now there is a coffee break, yes, which I'm sure we're all keen to get to. Thank you.
