Session
Description:
Generative AI is rapidly becoming a new intermediary for political information. Increasingly, citizens ask AI chatbots to explain party platforms, compare candidates, evaluate campaign promises, or even recommend how they should vote. Unlike search engines, large language models (LLMs) therefore no longer merely retrieve political information – they interpret it, prioritize it, and increasingly provide personalized voting advice. This development creates new opportunities for democratic participation while raising fundamental questions about information integrity, political pluralism, and electoral integrity. This workshop is designed to move from evidence to action. It begins with a discussion of the latest evidence on the provision of political information and voting advice by LLM-based chatbots – and public opinion on their use. Speakers will share original findings from a new 15-country public opinion survey on AI-generated voting advice (REMIT research project), alongside reports of chatbot use and regulation during recent elections in Brazil, Morocco and Nepal. These illustrate how AI-generated political advice is shaping democratic processes across different regulatory and political contexts. Building on this evidence, the discussion shifts toward governance. Representatives from different stakeholder groups will discuss what concrete safeguards are needed to ensure that AI-generated political advice supports democratic participation while protecting information integrity and political pluralism. They will also debate how human rights obligations and platform responsibilities can be viewed in the context of future multistakeholder governance of AI-generated political information and voting advice. The workshop builds on the intersessional work of the IGF Dynamic Coalition on Platform Responsibility and the IGF Dynamic Coalition on Internet Rights and Principles. It builds on the findings of previous IGF workshops, including “AI and Disinformation: Safeguarding Elections” #255, IGF 2024, and will be the next step toward a report published in 2027 as part of intersessional activities.
Generative AI is rapidly becoming a new intermediary for political information. Increasingly, citizens ask AI chatbots to explain party platforms, compare candidates, evaluate campaign promises, or even recommend how they should vote. Unlike search engines, large language models (LLMs) therefore no longer merely retrieve political information – they interpret it, prioritize it, and increasingly provide personalized voting advice. This development creates new opportunities for democratic participation while raising fundamental questions about information integrity, political pluralism, and electoral integrity. This workshop is designed to move from evidence to action. It begins with a discussion of the latest evidence on the provision of political information and voting advice by LLM-based chatbots – and public opinion on their use. Speakers will share original findings from a new 15-country public opinion survey on AI-generated voting advice (REMIT research project), alongside reports of chatbot use and regulation during recent elections in Brazil, Morocco and Nepal. These illustrate how AI-generated political advice is shaping democratic processes across different regulatory and political contexts. Building on this evidence, the discussion shifts toward governance. Representatives from different stakeholder groups will discuss what concrete safeguards are needed to ensure that AI-generated political advice supports democratic participation while protecting information integrity and political pluralism. They will also debate how human rights obligations and platform responsibilities can be viewed in the context of future multistakeholder governance of AI-generated political information and voting advice. The workshop builds on the intersessional work of the IGF Dynamic Coalition on Platform Responsibility and the IGF Dynamic Coalition on Internet Rights and Principles. It builds on the findings of previous IGF workshops, including “AI and Disinformation: Safeguarding Elections” #255, IGF 2024, and will be the next step toward a report published in 2027 as part of intersessional activities.
Policy Question(s)
1. What does current evidence tell us about the accuracy, neutrality, transparency, and political pluralism of voting advice generated by AI chatbots, and how are these perceived by users?
2. What responsibilities should AI providers, governments, election authorities, technical community and international organizations assume to ensure that AI-generated voting advice supports democratic participation and information integrity?
3. Which governance measures should be implemented at the platform, national, and international levels to ensure that AI-generated political advice is transparent, pluralistic, and consistent with democratic values and human rights?
Expected Outcomes
The workshop will directly contribute to a joint 2027 report of the IGF Dynamic Coalition on Platform Responsibility and the Dynamic Coalition on Internet Rights and Principles. This effort is open to all participants, both online and attending in person.The report will bring together evidence on current chatbot impact on political information and elections, and users' perceptions of AI-generated voting advice. Drawing on diverse regional and stakeholder perspectives, participants will develop tailored recommendations for policymakers, election authorities, AI developers, platforms, researchers, and civil society organizations. The report will be presented and discussed across National and Regional Initiatives (NRIs) and at IGF 2027, helping to inform future policy debates and governance approaches for AI-generated political information.
Format
Roundtable
Duration (minutes): 60
Format description: A 60-minute roundtable is well suited to the workshop's objective of moving from evidence to action. To maximize interaction, the session begins with an innovative 15-minute PechaKucha-style presentation (limited speaking time, enforced through automatically shifting and limited slides) in which speakers concisely present the latest evidence on AI-generated voting advice, public opinion about this, and regulatory experiences from different regions. This creates a shared evidence base while leaving sufficient time for moderated discussion among representatives from governments, the private sector, civil society, academia, and the technical community. The physical format (roundtable) encourages equal participation and collaborative problem-solving, while also engaging remote participants.
Duration (minutes): 60
Format description: A 60-minute roundtable is well suited to the workshop's objective of moving from evidence to action. To maximize interaction, the session begins with an innovative 15-minute PechaKucha-style presentation (limited speaking time, enforced through automatically shifting and limited slides) in which speakers concisely present the latest evidence on AI-generated voting advice, public opinion about this, and regulatory experiences from different regions. This creates a shared evidence base while leaving sufficient time for moderated discussion among representatives from governments, the private sector, civil society, academia, and the technical community. The physical format (roundtable) encourages equal participation and collaborative problem-solving, while also engaging remote participants.
Hybrid Format: The workshop is designed as a genuinely hybrid discussion rather than a sequence of presentations. A moderator will actively alternate between onsite and online speakers and participants, ensuring balanced participation throughout the session. Following the opening PechaKucha presentations, both onsite and remote participants will contribute equally during moderated discussion and stakeholder-specific recommendation building. Live polls (in-person and remote participants through online tool Mentimeter) will feature the same questions as those to which public opinion data presented thus making the evidence section of the workshop more engaging. Participants will also be able to submit questions through the and chat, with a dedicated online moderator ensuring that remote interventions receive equal attention.
Organizer 1: Yasmin Curzi de Mendonça, UVA Karsh Institute of Democracy
Organizer 2: Dennis Redeker, University of Bremen
Organizer 3: Santosh Babu Sigdel, 🔒Digital Rights Nepal
Organizer 4: Adio Dinika, University of Bremen
Organizer 2: Dennis Redeker, University of Bremen
Organizer 3: Santosh Babu Sigdel, 🔒Digital Rights Nepal
Organizer 4: Adio Dinika, University of Bremen
Speaker 1: Yasmin Curzi de Mendonça, Technical Community, Latin American and Caribbean Group (GRULAC)
Speaker 2: Ana Ruelas, Intergovernmental Organization, Intergovernmental Organization
Speaker 3: Adio Dinika, Technical Community, African Group
Speaker 4: Ojo Edetaen, Civil Society, African Group
Speaker 2: Ana Ruelas, Intergovernmental Organization, Intergovernmental Organization
Speaker 3: Adio Dinika, Technical Community, African Group
Speaker 4: Ojo Edetaen, Civil Society, African Group
Moderator
Adio Dinika, Technical Community, African Group
Online Moderator
Dennis Redeker, Civil Society, Western European and Others Group (WEOG)
Rapporteur
Santosh Babu Sigdel, Civil Society, Asia-Pacific Group
What will participants gain from attending this session?
