Artificial Intelligence and the Future of Citizen Participation

AI tools offer significant opportunities to support policymakers and practitioners in designing and implementing meaningful citizen participation processes. Government should seize these opportunities while acting to mitigate the numerous risks entailed by the adoption of AI. In parallel, governments continue working to improve and embed meaningful citizen participation practices in policymaking, beyond the use of digital technologies. Table 4.1 provides a summary of the mitigation strategies to ensure trustworthy uptake of AI tools for participation. These strategies fall under the broader scope of the OECD Framework for Trustworthy Artificial Intelligence in Government (OECD, 2025[1]) which recommends governments to adopt robust Guardrails to counter risks, strengthen Enablers to facilitate use and overcome implementation challenges, and promote Engagement of citizens and stakeholders in shaping and regulating fit-for-purpose AI systems. In addition, the OECD GovTech Policy Framework, as well as the Bertelsmann Stiftung's Public AI Whitepaper present comprehensive integrated approaches for governments to respectively succeed in their digital transformation and steer safe AI ecosystems that build on digital public infrastructure, including in collaboration with non-governmental entities (OECD, 2024[2]; Bertelsmann Stiftung, 2025[3]; OECD, 2024[4]).

To prevent adverse outcomes, governments should be cautious in the choice of AI tools they seek to adopt, making sure that the underlying models comply with the OECD AI Principles, including principles of respect for the rule of law, human rights, and democratic values, including fairness and privacy (OECD, 2024[5]).

For instance, governments can act to close data divides. To address the language divide, the government of Iceland partnered with OpenAI to train the Large Language Model (LLM) GPT-4 in Icelandic in order to preserve the Icelandic language (Government of Iceland, 2023[6]). Similarly, the University of Turku (Finland) partnered in 2023 with the company SiloAI to build the Poro model, a family of multilingual open-source LLMs for all European official languages. Poro runs on LUMI, one of the nine supercomputers of the EU High Performance Computing Joint Undertaking (University of Turku, 2023[7]).

Governments can also act to address privacy concerns by establishing clear frameworks, guidelines, and mechanisms to promote strong data governance that safeguards privacy, intellectual property, and security. Countries like Korea, New Zealand, and the UK have developed dedicated guidance for safe and privacy-preserving use of AI in government, personal information processing and AI development, while France established a data sandbox to create space for safe experimentation (OECD, 2025[1]).

The OECD AI Principles ascribed transparency and explainability to the core principles of AI systems’ development and governance (OECD, 2024[5]). To promote transparency and auditability of AI, the government of the Netherlands launched in 2022 an Algorithm Register to provide information about the algorithms used by the Dutch Government and guidance to object an algorithm-based decision (Government of the Netherlands, 2022[8]).

To improve the transparency of digital tools designed and used for citizen participation processes, the civic tech ecosystem adopted and advocated for open-source software, meaning “software the users have the freedom to run, copy, distribute, study, change and improve” (OECD, forthcoming[9]; The Open Source Initiative, 2023[10]). According to People Powered’s 2025 rating of digital platforms for citizen participation, 10 out of the 10 best ranked platforms are open source (People Powered, 2025[11]). The open-source paradigm is more complex to apply to AI systems, as it involves multiple layers underpinning the technology: source code, documentation, training data or pre-trained model weights1. Depending on the system, the label of open-source AI might then be misleading (OECD, 2025[12]). Alternative paradigms of transparency of AI systems include open standards, i.e., technical specifications or protocols that are developed through transparent processes and are publicly available for anyone to use, implement, or improve (Almeida, Oliveira and Cruz, 2011[13]; Zielke, 2020[14]; Mourne, O’Reilly and Strauss, 2025[15]), and, in the case of foundation models, open weights, i.e. publicly available trained weights (OECD, 2025[12]).

Moreover, the use of AI tools for citizen participation should be directed towards enhancing governments’ capacity to listen to all citizens. To prevent the exclusion of citizens with inadequate skills or limited access digital and AI tools, governments should provide alternative in-person or low-tech formats of participation (OECD, 2022[16]). The efficiency gains allowed AI tools should serve to improve resource allocation towards these alternative formats. Governments should also invest to strengthen citizens’ AI literacy. In April 2025, the US issued an executive order mandating a task force to elaborate an action plan to foster AI education among youth (The White House, 2025[17]).

Governments should also be open about their use of AI tools as part of citizen participation processes, to ensure process transparency, and establish clear redressal mechanisms to address citizens’ grievances. Furthermore, to prevent public resistance and strengthen public trust, governments should be careful in their use of AI tools in all policy functions, and particularly those where misuses of AI could undermine the civic space. Regulations, ethical frameworks, and dedicated strategies can align practices with the respect of civil freedoms.

Governments can address the existing gap in awareness and guidance on opportunities and risks of AI in citizen participation. The typology presented in this report, together with the OECD Policy Paper “Tackling Civic Participation Challenges with Emerging Technologies: Beyond the Hype” (OECD, 2025[18]) can constitute useful resources for governments seeking such guidance.

Awareness initiatives and training opportunities for civil servants are key to preventing these risks and allow governments to unlock the full potential of AI tools to enhance citizen participation and strengthen public trust. For instance, the European Union developed a comprehensive competency framework for AI in government (Medaglia, Mikalef and Tangi, 24[19]), comprising three dimensions (technology, managerial, and policy, legal and ethical) and three cross-cutting clusters (attitudinal competences, operational competences, and literacy competences).

Beyond guidelines and best practices, governments can map and strengthen innovation ecosystems at the intersection of AI development and citizen participation (OECD, 2025[20]), and build intergovernmental, international, and multi-stakeholder co-creation platforms, innovation labs, and communities of practice (OECD, 2024[2]). In addition to the OECD Innovative Citizen Participation Network, the OECD is currently building a Community of Practice on Emerging Technologies for Civic Participation. Similarly, the Bertelsmann Stiftung is co-founder of the Network on Citizen Participation and Deliberation in Europe.

Beyond experimentations and pilots, governments should pursue coordinated approaches that enable the adoption of AI tools at scale and their dedicated use for citizen participation. For example, Sitra’s (Finland) pilots with the platform Polis are part of a broader strategy to equip government at the national and local levels with tools and skills to better interact with citizens (Lovio, 2025[21]) In the United Kingdom, the iAI Incubator is creating a suite of open-source AI tools for all levels and branches of the British public administrations, including Consult (see Box 2.8) (AI.GOV.UK, 2025[22]).

Finally, governments can strengthen their capacity to procure AI solutions to foster innovation that is fit-for-purpose (Monteiro, Hlacs and Boéchat, 2024[23]). Procurement processes of AI tools should carefully consider the consequences of potential vendor and data lock-ins and opt for systems that guarantee interoperability (OECD, 2024[2]).

AI systems have the potential to radically reshape the interaction between citizens themselves, and between citizens and their governments. There are several benefits in involving citizens in the design and development of AI-based technologies used by governments: greater trust and legitimacy of AI systems and stronger reflection of the needs of users (OECD, 2024[24]). Consistently with the provisions of the Engagement pillar of the OECD Framework on Trustworthy AI in Government, citizens and stakeholders can be involved at different stages of the technology cycle (OECD, 2025[1]). For example, in 2023, the United Kingdom engaged in a series of public consultations with more than 300 stakeholders to build a regulatory framework for AI to drive safe, responsible innovation (United Kingdom Government, 2024[25]). Similarly, in 2024 the government of Brazil involved more than 300 citizens and stakeholders to shape its AI plan (Government of Brazil, 2024[26]).

Citizens can contribute to the definition of priorities and principles guiding the development and governance of AI in government and in society. Representative deliberative processes, which are meant to address the type of policy issues that have long-term implications and carry society-wide value-based dilemmas, requiring trade-offs, are particularly well suited to dealing with AI policy and governance decisions (OECD, 2020[27]; Hintz, 2021[28]). AI involves ethical and societal discussions to decide on its uses in specific contexts (e.g. facial recognition, online harm). Most importantly, the adoption of AI technologies can shape social interactions in the long term, with impacts that can span several generations. In 2024, the Belgian Presidency of the Council of the European Union launched a Citizen Panel gathered a representative group of 60 Belgians to collect their views on artificial intelligence and the role of the EU in developing and regulating it (Belgian Presidency, 2024[29]). Similarly, in late 2024 the Citizens’ Jury “AI and Freedom” of the German state of Baden-Württemberg formulated nine recommendations on how citizens can be involved in the publicly funded research and development of AI. The members of the mini-public presented their findings to Science Minister Petra Olschowski on 10 March 2025. Citizen assemblies on AI could be relevant and impactful at all levels of governance, from local (Verhulst and Chwalisz, 2025[30]) to global (Davies et al., 2024[31]).

Governments can harness citizen and stakeholder expertise to shape AI tools and models that meet societal needs through citizen participation processes such as open innovation and co-creation. For example, in 2020 the French Government launched PIAF, a collaborative initiative with citizens, academia and civil society to build databases in French language to train AI models. In 2023, the tech company Anthropic, which developed the AI-model Claude, partnered with the Collective Intelligence Project to involve a representative sample of ~1000 American citizens in the collaborative drafting of a constitution for an AI system using Polis (Ganguli and al., 2023[32]). In 2024, the government of India organised an open online hackathon on data-driven innovation to address citizens grievances. The five problem statements revolved around the development of AI-driven systems to cluster and label citizen grievances, create virtual assistants for both citizens and civil servants, and improve the accuracy of speech-to-text transcriptions of feedback calls in Hindi and English (Innovate India, 2024[33]). Between 2023 and 2025, Portugal, Spain and the Netherlands worked together to co-create AI systems that could address real-life challenges of citizen participation (see Box 4.1).

Citizens and stakeholders have a fundamental role to play in the oversight of AI tools. For instance, in the United States, the Expert & Citizen Assessment of Science & Technology (ECAST) Participatory Technology Assessment is bringing public perspectives to discuss critical government science and technology decisions (Weller, Farooque and Sullivan Govani, 2021[35]). Citizens, researchers, and civil society organisations are also challenging AI models by applying Red Teaming techniques, which use adversarial methods to identify vulnerabilities in systems (UNESCO, 2025[36]; Singh et al., 2025[37]).

International AI governance could benefit from broader input, including that of citizens and stakeholders. Lessons can be drawn by the precedent of the world wide web. The architecture of global governance for the Internet is based on a multi-stakeholder process where public authorities and non-governmental actors have an equal voice (see Box 4.2). Although this model does not come without limitations, including power imbalances and lack of binding authority, the establishment of multi-stakeholder fora where citizens are represented can ensure multiple voices are heard and can influence how technologies are shaped. In Brazil, national institution, academia, private sector provider, and third sector advocates converge in the Brazilian Internet Steering Committee to establishing strategic guidelines related to the use and development of the Internet in Brazil, such as the allocation of IP Addresses and the administration of the Domain (CGI.br, 2025[38]).

The OECD’s Global Partnership for Artificial Intelligence (GPAI), comprising 44 member countries, involves a multidisciplinary community of more than 500 AI experts from government, industry, academia, and civil society. The latter is represented through the Civil Society Information Society Advisory Council (CSISAC) (OECD.AI, 2025[40]; CSISAC, 2025[41]).

[22] AI.GOV.UK (2025), Projects, https://ai.gov.uk/projects/.

[13] Almeida, F., J. Oliveira and J. Cruz (2011), “Open standards and open source: enabling interoperability.”, International Journal of Software Engineering & Applications, Vol. 2/1.

[29] Belgian Presidency (2024), A Citizen’s View of Artificial Intelligence within the EU, https://demnet.hu/wp-content/uploads/2024/05/rapport-ia-en-web-v2.pdf.

[3] Bertelsmann Stiftung (2025), Public AI Whitepaper, https://doi.org/10.11586/2025040.

[38] CGI.br (2025), Sobre o CGI.br, https://cgi.br/sobre/.

[41] CSISAC (2025), , https://csisac.org/about.php.

[31] Davies, T. et al. (2024), Global Citizen Deliberation on Artificial Intelligence: Options and design considerations, https://connectedbydata.org/projects/2024-gca-ai.

[32] Ganguli, D. and A. al. (2023), Collective Constitutional AI: aligning a language model with public input, https://www-cdn.anthropic.com/b43359be43cabdbe3a8ffd60ea8a68acf25cb22e/Anthropic_CollectiveConstitutionalAI.pdf.

[26] Government of Brazil (2024), IA para o bem de todos, https://www.gov.br/mcti/pt-br/acompanhe-o-mcti/noticias/2024/07/plano-brasileiro-de-ia-tera-supercomputador-e-investimento-de-r-23-bilhoes-em-quatro-anos/ia_para_o_bem_de_todos.pdf/view.

[6] Government of Iceland (2023), Head Start for Icelandic, https://www.government.is/diplomatic-missions/embassy-article/2023/03/14/Head-start-for-Icelandic/ (accessed on 23 May 2024).

[8] Government of the Netherlands (2022), The Algorithm Register of the Dutch Government, https://algoritmes.overheid.nl/en (accessed on 23 May 2024).

[28] Hintz (2021), TOWARDS CIVIC PARTICIPATION IN THE DATAFIED SOCIETY: can citizen assemblies democratise algorithmic governance?, https://doi.org/10.5210/spir.v2021i0.11943.

[39] ICANN (2024), How does ICANN work?, https://www.icann.org/resources/pages/welcome-2012-02-25-en (accessed on 23 May 2024).

[33] Innovate India (2024), Online Hackathon on Data-driven Innovation for Citizen Grievance Redressal-2024, https://innovateindia.mygov.in/darpg-challenge-2024/.

[34] LabX - ARTE (2025), Improving civic participation through emerging technologies, https://labx.gov.pt/projetos-posts/improving-civic-participation-through-emerging-technologies/?lang=en.

[21] Lovio, I. (2025), Sitra to significantly increase funding for solutions that renew Finland, https://www.sitra.fi/en/news/sitra-to-significantly-increase-funding-for-solutions-that-renew-finland/.

[19] Medaglia, R., P. Mikalef and L. Tangi (24), Competences and governance practices for artificial intelligence in the public sector, Publications Office of the European Union, https://doi.org/10.2760/7895569.

[23] Monteiro, B., A. Hlacs and P. Boéchat (2024), “Public procurement for public sector innovation: Facilitating innovators’ access to innovation procurement”, OECD Working Papers on Public Governance, No. 80, OECD Publishing, Paris, https://doi.org/10.1787/9aad76b7-en.

[15] Mourne, I., T. O’Reilly and I. Strauss (2025), Open Protocols Can Prevent AI Monopolies, https://ai-frontiers.org/articles/open-protocols-prevent-ai-monopolies.

[12] OECD (2025), AI openness: A primer for policymakers, OECD Publishing, https://doi.org/10.1787/02f73362-en.

[1] OECD (2025), Governing with Artificial Intelligence: The state of play and way forward in core government functions, OECD Publishing.

[20] OECD (2025), How Innovation Ecosystems Foster Citizen Participation Using Emerging Technologies in Portugal, Spain and the Netherlands, OECD Publishing, https://doi.org/10.1787/2cb37a30-en.

[18] OECD (2025), “Tackling civic participation challenges with emerging technologies: Beyond the hype”, OECD Public Governance Policy Papers, No. 72, OECD Publishing, Paris, https://doi.org/10.1787/ec2ca9a2-en.

[4] OECD (2024), “Digital public infrastructure for digital governments”, OECD Public Governance Policy Papers, No. 68, OECD Publishing, Paris, https://doi.org/10.1787/ff525dc8-en.

[2] OECD (2024), Enabling Digital Innovation in Government: The OECD GovTech Policy Framework, OECD Digital Government Studies, OECD Publishing, Paris, https://doi.org/10.1787/a51eb9b2-en.

[24] OECD (2024), “Framework for Anticipatory Governance of Emerging Technologies”, OECD Science, Technology and Industry Policy Papers, No. 165, OECD Publishing, Paris, https://doi.org/10.1787/0248ead5-en.

[5] OECD (2024), Recommendation of the Council on Artificial Intelligence, https://legalinstruments.oecd.org/en/instruments/oecd-legal-0449.

[16] OECD (2022), OECD Guidelines for Citizen Participation Processes, OECD Public Governance Reviews, OECD Publishing, Paris, https://doi.org/10.1787/f765caf6-en.

[27] OECD (2020), Innovative Citizen Participation and New Democratic Institutions: Catching the Deliberative Wave, OECD Publishing, https://doi.org/10.1787/339306da-en.

[9] OECD (forthcoming), Background Note: Getting Civic Tech Right for Democracy.

[40] OECD.AI (2025), About the Global Partnership on Artificial Intelligence (GPAI), https://oecd.ai/en/about/about-gpai.

[11] People Powered (2025), 2025 Rating for Digital Participation Tools, https://www.peoplepowered.org/platform-ratings.

[42] Seger, E. et al. (2023), “Open-Sourcing Highly Capable Foundation Models: An evaluation of risks, benefits, and alternative methods for pursuing open-source objectives”, LawAI Working Paper, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4596436.

[37] Singh, R. et al. (2025), Red-Teaming in the Public Interest, https://ranjitsingh.me/wp-content/uploads/2025/02/Singh_Red-Teaming-in_the_Public_Interest.pdf.

[10] The Open Source Initiative (2023), The Open Source Definition, https://opensource.org/osd/.

[17] The White House (2025), ADVANCING ARTIFICIAL INTELLIGENCE EDUCATION FOR AMERICAN YOUTH, https://www.whitehouse.gov/presidential-actions/2025/04/advancing-artificial-intelligence-education-for-american-youth/.

[36] UNESCO (2025), Red Teaming Artificial Intelligence for Social Good - The Playbook, https://unesdoc.unesco.org/ark:/48223/pf0000394338.locale=en.

[25] United Kingdom Government (2024), A pro-innovation approach to AI regulation: government response, https://www.gov.uk/government/consultations/ai-regulation-a-pro-innovation-approach-policy-proposals/outcome/a-pro-innovation-approach-to-ai-regulation-government-response#introduction.

[7] University of Turku (2023), TurkuNLP, https://turkunlp.org/ (accessed on 23 May 2024).

[30] Verhulst, S. and C. Chwalisz (2025), The case for local and regional public, https://www.demnext.org/uploads/AI-Localism-paper.pdf.

[35] Weller, N., M. Farooque and M. Sullivan Govani (2021), “Need Public Policy for Human Gene Editing, Heatwaves, or Asteroids? Try Thinking Like a Citizen”, Perspectives, Vol. XXXVII/3, https://issues.org/thinking-like-citizen-participatory-technology-assessment-weller-govani-farooque/.

[14] Zielke, T. (2020), “Is Artificial Intelligence Ready for Standardization?”, Communications in Computer and Information Science, Vol. 1251, https://doi.org/10.1007/978-3-030-56441-4_19.