Policy Brief: The evolving role of trust in AI-Mediated Societies

Knowledge T echnologies for Democracy POLICY BRIEF December 2025 DOI Number: 10.5281/zenodo.17975874 The Evolving Role of Trust in AI-Mediated Societies Authors Trinity College Dublin Jennifer Edmond Fundación Cibervoluntarios María Ruíz de Assin de los Santos Demos Helsinki Atte Ojanen Strane Innovation Tiffany Morisseau Warwick University Jessica Sutherland Trust-IT Services Gabriela Rogowska Gabriele Quattrocchi Graphic Design Gianluca Savini e Andrea Greco - Trust-IT Services Organisations involved in KT4D The Evolving Role of Trust in AI-Mediated Societies 1 Table of contents Executive Summary ...................................................................................................................................................2 1. AI as a new actor in the trust equation .............................................................................................................3 1.1 Societal Challenges ..........................................................................................................................................................3 1.2 Policy Challenges ...........................................................................................................................................................5 2. Recommendations ................................................................................................................................................6 Recommendation 1: Ensure the traceability and authenticity of information by public institutions ....6 Recommendation 2: Encourage organisations to communicate on shared rules of AI use to sustain social trust.................................................................................................................................................................6 Recommendation 3: Foster contextualised critical digital literacy programmes ........................................7 Recommendation 4: Ensure AI governance is responsive to interculturality and cultural differences 8 Recommendation 5: Establish mandatory human oversight for all AI systems involved in critical decision-making, particularly in employment, education, healthcare, and legal contexts. ......................9 Recommendation 6: Ensure epistemic responsibility in online knowledge dissemination through increased scrutiny and regulation of online platforms .........................................................................................10 Recommendation 7: Adopt broader, infrastructural lens to democratic AI governance .........................10 Recommendation 8: Ensure sufficient capacities to enforce current AI regulation ..................................10 3. About KT4D Project ........................................................................................................................................... 11 The Evolving Role of Trust in AI-Mediated Societies 2Executive Summary Trust in democratic institutions is being fundamentally redefined as Artificial Intelligence increasingly mediates information exchange and influences decision-making processes. The KT4D project examines how artificial intelligence affects trust at multiple levels: between citizens and institutions, across different cultural contexts, and within organisations where Artificial Intelligence is deployed without clear standards. The policy brief presents eight recommendations grouped across institutional, organisational, educational, and governance dimensions. At the institutional level, public bodies should ensure traceability and authenticity of information through authenticated digital signatures and origin labels. Organisations deploying AI systems must establish and communicate shared rules for AI use to keep interpersonal and institutional trust. Educational recommendations emphasise contextualised critical digital literacy programmes that reflect local languages, social norms, and cultural values. Governance recommendations call for AI frameworks responsive to intercultural differences, mandatory human oversight for critical decisionmaking systems, epistemic responsibility in online knowledge dissemination, infrastructural approaches to democratic AI governance, and sufficient enforcement capacities for current regulation. The central policy challenge is that while the European Union has addressed the effects of artificial intelligence through regulations, the focus has recently shifted toward innovation and deregulation. This creates tension between upholding democratic processes and promoting the rapid deployment of artificial intelligence, as this advanced knowledge technology can indirectly disrupt democracy in ways that static risk categories fail to capture. The Evolving Role of Trust in AI-Mediated Societies 3 1. AI as a new actor in the trust equation 1 Edelman. (2025). 2025 Edelman Trust Barometer. Edelman. https://www.edelman.com/trust Trust is essential in democratic societies, forming a precondition for citizens to rely on institutions and engage in decision-making. Yet, trust is being redefined as Artificial Intelligence (AI) increasingly mediates information exchange, and affects decisionmaking processes and public discourses. The influence of algorithms and the rapid dissemination of news and information across cultural contexts make it increasingly difficult to identify trustworthy sources of knowledge. As of today, trust depends not only on institutional integrity but also on digital literacy and technical transparency. Understanding how trust is built, maintained, and sometimes eroded in technologically mediated environments is central to safeguarding critical thinking and democratic participation. 1.1 Societal Challenges What does it mean to receive, understand, or produce information in AI-mediated environments? The growing likelihood that any given piece of information, image or video has been generated or transformed by AI obfuscates how citizens assess and relate to information. It is becoming increasingly complex to identify in what context information was produced and the intentions underpinning it. The uncertainty caused by this opacity can (sometimes intentionally) undermine citizens’ trust in institutions, which are less able to serve as reliable reference points. Therefore, maintaining trust depends on strong collective mechanisms ensuring that information shared by institutions can be reliably traced, verified, and contextualised, especially on issues that truly matter to citizens. The intertwining of trust in institutions and trust in AI The 2025 Edelman Trust Barometer1 shows a clear trend: people who feel a stronger sense of grievance — those who believe that the system has failed or excluded them — tend to trust artificial intelligence less and feel less comfortable with its use by businesses. This decline in confidence reflects more than technological skepticism; it points to a deeper decay of institutional and social trust. This helps explain the 22-point drop in AI trust among those with high grievance in the survey. For many, the technology is not viewed as a tool for transparency, but rather as another source of informational uncertainty. The Evolving Role of Trust in AI-Mediated Societies 4 Figure 1. The figure adopted from the Edelman Trust Barometer (2025). The role of culture in trust mediation Attitudes toward AI and institutional trust are also culturally mediated. Depending on whether a country has high institutional trust or not, citizens may view AI as a tool for efficiency and public benefit, or exclusion and manipulation. There is also a clear trust gap between the Global North and South. Citizens in Western democracies, such as the US, UK, Germany, and Japan, generally show low and declining trust in AI (37–47%). In contrast, many countries in the Global South, particularly in Asia and the Middle East (China, India, UAE, Indonesia, Saudi Arabia), exhibit high and stable trust in the technology (70–77%). This pattern suggests that trust is not merely about elections or democratic participation but shaped by broader structural, social, cultural, and governance factors, creating a North–South divide. Trust within organisations Another challenge concerns how AI affects trust within organisations. The development of AI in collective and institutional settings has indeed direct effects on interpersonal trust between members of communities, whether corporate, scientific, or educational.. Beyond individual interactions, AI affects how these communities negotiate shared norms. When the use of AI is introduced without clear standards, it can generate uncertainty not only linked to the technology itself, but also about people’s intentions, competencies and responsibilities. The absence of a shared understanding of what is an acceptable or legitimate use of AI may erode the mutual expectations underlying cooperation. While there are some generally shared intuitions on this issue (for example, the use of AI is considered more legitimate for administrative support than hiring decisions), establishing explicit and acceptable frameworks for the use of AI is therefore a social and democratic necessity. (For related insights, see KT4D Social Risk Toolkit Module B: AI, trust and awareness, “AI and the valuation of human work” DOI 10.5281/zenodo.17977403). The Evolving Role of Trust in AI-Mediated Societies 5 1.2 Policy Challenges 2 Brussels, 19.11.2025 COM(2025) 836 final 2025/0359 (COD) Proposal for a Regulation Of The European Parliament and of the Council amending Regulations (EU) 2024/1689 and (EU) 2018/1139 as regards the simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI). 3 Brussels, 9.4.2025 COM(2025) 165 final Communication From The Commission To The European Parliament, The Council, The European Economic And Social Committee And The Committee Of The Regions, AI Continent Action Plan. The European Union (EU) is tackling AI’s societal effects through a multifaceted set of policies, with implications for institutional trust. On the regulatory front, laws like the GDPR, the Digital Services Act (DSA), and the Digital Markets Act (DMA) govern data processing and the behaviour of major online platforms where disinformation often spreads. The EU AI Act (Regulation (EU) 2024/1689) addresses democratic threats more directly by classifying systems that influence voting as “high-risk,” while new democracy-focused policies like the European Democracy Shield [Brussels, 12.11.2025 JOIN(2025) 791 final - European Democracy Shield: Empowering Strong and Resilient Democracies] are designed to combat foreign information manipulation and protect media pluralism. The central policy challenge, however, is that AI can also indirectly disrupt democracy in ways that static risk categories fail to capture. At the same time, the EU focus has shifted towards innovation and deregulation, as evident in the Digital Omnibus2 and the AI Continent Action Plan3. The latter includes initiatives like the Apply AI Strategy, the Cloud and AI Development Act, and the development of AI Factories. While digital sovereignty and public AI infrastructure is certainly needed to build trust, one can also identify a tension between upholding democratic processes and promoting rapid, profit-driven AI deployment across society. Figure 2. Overview of the AI-related policies. The Evolving Role of Trust in AI-Mediated Societies 62. Recommendations ✅ Recommendation 1: Ensure the traceability and authenticity of information by public institutions Public institutions should establish systems that guarantee the traceability and authenticity of the information they produce and disseminate (for example, through authenticated digital signatures or institutional origin labels certifying that the content comes from an identifiable public body). These mechanisms should include a visible and easily understandable sign or label, allowing citizens to recognise genuine institutional communication.  Why it is important As AI-generated content is growing exponentially, citizens face growing uncertainty about the origin and reliability of information. Ensuring that information coming from public bodies can be clearly identified and traced back to their legitimate source is therefore essential to maintain trust and to prevent the manipulation of public discourse. ✅ Recommendation 2: Encourage organisations to communicate on shared rules of AI use to sustain social trust Each organisation that deploys AI systems (whether in public administration, research, education, business, or civil society) should establish and make available its own shared rules for AI use, such as universities’ guidelines for the use of generative AI. Ideally these are discussed internally and informed by a clear mapping of how AI is used across the organisation’s activities.  Why it is important When organisations adopt AI tools without clear internal standards, uncertainty arises not only about the technology but also about human intentions, competencies, and responsibilities. Collaborators and citizens need to know when and how AI is expected to be used. Without this shared understanding, interpersonal and institutional trust may be undermined. The Evolving Role of Trust in AI-Mediated Societies 7 ✅ Recommendation 3: Foster contextualised critical digital literacy programmes Design and implement contextual critical digital literacy programmes within local institutions (schools, libraries, and civil society organisations) that integrate and embed social, linguistic, and political specificities of each community. These should include critical consumption skills, including analysis and awareness.  Why it is important Digital and AI literacy cannot be universalised; it must reflect local languages, social norms, and cultural values. A culturally contextualised approach ensures that citizens understand and trust the digital tools that shape democratic participation. Figure 3. Illustration of the components of critical digital literacy: awareness, trust and vulnerability