Why the EU’s Technosolutionist Focus on AI and Media ‘Literacy’ Empowers Big Tech: Centering Structural Approaches to Counter the Undemocratic Political Economy of Surveillance Capitalism

Abstract

The emergence of digital technologies during the last two decades has placed strain on democracies globally. From disinformation to artificial intelligence (AI), policy-makers have struggled to address the authoritarian and anti-democratic tendencies that Big Tech companies have been pushing. A dominant response from European Union (EU) policy-makers has been to promote ‘literacy’: media literacy to address disinformation, and ‘AI literacy’ to foster constructive uses of AI. We ask: what does it mean for the EU to use media literacy and AI literacy as a response to disinformation and the risks of AI? More broadly, what kind of policy and model of democracy is being constructed when the EU suggests that the solution to disinformation and AI depends on citizens becoming more “literate”? We empirically examine the usage of ‘literacy’ in EU policy documents in the context of disinformation and AI, and argue that it shifts responsibility from platforms to individual citizens. In doing so, it moves attention away from a structural approach into the political economy of Big Tech companies, hence empowering them and their ‘surveillance capitalist’ business model. This article argues instead that literacy needs to be rethought beyond individual skills, as a way of making visible the structural and infrastructural power of privately owned digital systems shaping contemporary public spheres. Therefore, reclaiming democracy requires moving beyond individual adaptation, and instead towards addressing the underlying systemic structures of Big Tech power.

1. Introduction: Literacy as a ‘Solution’ to Tackle Systemic Problems Such as Disinformation and AI

The emergence of digital technologies over the past two decades has shaped how democracies worldwide operate. From disinformation to artificial intelligence (AI), policy-makers globally have struggled to address the authoritarian and anti-democratic trends that have accompanied the consolidation of power by Big Tech companies such as Google, Meta, Apple, Microsoft or Amazon, and more recently OpenAI and Anthropic. In this context, institutional responses have not emerged in isolation, but as part of a broader attempt to democratically stabilise increasingly complex digital environments, which has progressively led to the adoption of approaches centred on individual citizen adaptation. Indeed, a salient response from policymakers, particularly in the context of the European Union (EU), has been to promote ‘literacy’: media literacy to tackle disinformation, and ‘AI literacy’ to promote the responsible use and social integration of AI systems. In this article, we empirically examine the usage of ‘literacy’ in EU policy documents in the context of disinformation and AI, and critically analyse what this ‘literacy’ approach entails for tackling disinformation and AI, focusing on the case of the EU, which is often presented as the most advanced regulator of Big Tech companies (Bradford 2023). We ask: what does it mean for the EU to use media literacy and AI literacy as a response to disinformation and the risks of AI? More broadly, what kind of policy and model of democracy is being constructed when the EU suggests that the solution to disinformation and AI depends on citizens becoming more “literate”?

This article contributes to the literature on platform governance, technosolutionism and digital democracy by conceptualising media literacy and AI literacy not only as educational tools, but also as governance mechanisms that shape how responsibility for digital harms is distributed among platforms, institutions and citizens. The existing literature has critically examined how private actors (including Big Tech companies) have influenced online safety education and promote individualising narratives (Estellés and Doyle 2025). In the EU context, colleagues have looked at how the European Commission has progressively moved towards emphasising ‘skills’ as a key employment strategy, emphasising individual responsibility (see Crespy and Kenn 2024). However, less attention has been paid to the specific role of media literacy and AI literacy in the specific context of the EU and its digital policies. Hence, there is a research gap in tracing how literacy functions more broadly as a technology of governance that redistributes political responsibility from infrastructures and corporate actors towards individuals, thereby helping to manage digital risks without necessarily addressing their structural causes and the political economy of technology.

Building on this research gap, the article illustrates that these policies do not operate solely as neutral educational strategies, but have deeper political effects. They frame the problems of disinformation and AI as primarily individual issues, reduce regulatory pressure on platforms, and contribute to stabilising the economic model of platform and surveillance capitalism (Zuboff 2019). Media literacy and AI literacy do not only fail to correct the structural power dynamics of Big Tech, but actually contribute to make them more socially and politically acceptable. The argument we present in the paper is not a direct causal effect, but rather an argument about the legitimisation and normalisation of Big Tech by the EU, where responsibility is increasingly shifted towards individuals (often conceived as ‘users’ rather than citizens) rather than centering structural approaches to counter the undemocratic political economy of surveillance capitalism. In this sense, we distinguish between the democratic potential of media literacy—even if we are critical about its specific mobilisation in EU policy documents—and the normalisation of AI embedded in the notion of AI literacy.

From a methodological perspective, the article combines elements of critical discourse analysis, and interpretative public policy analysis, applied to EU institutional and political documents relating to disinformation, media literacy, AI governance, and AI literacy between 2018 and 2026. The analysis focuses on how democratic risks are framed, how responsibility is distributed among actors, and what forms of political subjectivity are constructed within the discourse of EU digital governance. Following this introduction, Section 2 puts forward the theoretical framework surrounding technosolutionism, platform governance and literacy as a mechanism of governance. Section 3 presents the methodological approach and the selection of the corpus. Section 4, first, analyses the EU’s media literacy approach to countering disinformation, while the second part focuses on AI literacy and the normalisation of algorithmic governance. Section 5 develops a broader discussion of the relationship between literacy, technosolutionism and the political economy of Big Tech. Finally, the conclusion summarises the main findings and reflects on the implications of addressing systemic democratic problems through predominantly individualised policy responses.

2. The Political Economy of Big Tech and the EU: “Literacy” as a Technosolutionist Governance Approach to Disinformation and AI

Over the last decade, the EU has established itself as one of the most active players in the regulation of digital platforms, disinformation and, more recently, AI. This regulatory shift intensified particularly following events such as the 2018 Cambridge Analytica scandal and the growing debate on the democratic impact of large digital platforms, helping to reinforce the EU’s image as a pioneer in global technology governance and as the leading exponent of the so-called ‘Brussels Effect’ (Bradford 2020, 2023). In addition to the 2016 General Data Protection Regulation (GDPR), between 2022 and 2024, the EU has driven an unprecedented regulatory agenda, particularly through the Digital Services Act (DSA), the Digital Markets Act (DMA), the AI Act and the European Media Freedom Act (EMFA). This intense regulatory activity has been accompanied by strong lobbying efforts by Big Tech companies, with an important degree of success (Bouza García and Oleart 2024; Bouza et al. 2025).

Within EU regulatory frameworks, media literacy and, more recently, AI literacy have taken on an increasingly central role. There have been multiple strategies to tackle disinformation (see Farkas and Schou 2019), among which media literacy, fact-checking, strategic communication and digital literacy are often highlighted in the EU context (see Tuñón Navarro et al. 2019, 2025). Ördén (2019, p. 422) identifies media literacy as one of the most recurrent responses within European frameworks for disinformation governance, alongside strategic communication and media pluralism. The literature on the political economy of Big Tech companies has long identified Big Tech’s business model as a central challenge for contemporary democracies, particularly in relation to the circulation of disinformation and the deployment of AI systems (Srnicek 2016; van Dijck 2020). Digital platforms are increasingly understood as being structured around systems of large-scale data collection, computational modelling and algorithmic optimisation of communication, which enhance the capacity for individualised targeting (resulting often in targeted advertisement) and create structural conditions (Tufekci 2014) that can intensify emotionally charged, divisive or highly shareable forms of content because it increases ‘user engagement’. In this context, disinformation and the expansion of algorithmic systems do not appear as isolated phenomena, but as dynamics closely linked to the very economic and technological architecture of surveillance capitalism (see Zuboff 2019). Disinformation is therefore the logical result of social media companies’ attempt at maximising engagement (and hence their revenues) in their platforms.

Beyond its economic dimension, this architecture also entails new forms of infrastructural dependency between public and private actors. As van Dijck (2020) points out, large platforms have ceased to function solely as technological intermediaries and have become central infrastructures for communication, the circulation of information and digital governance. This creates a situation in which public institutions are increasingly dependent on private systems to manage democratic risks, moderate content or coordinate responses to disinformation and AI. Within this framework, digital literacy emerges as a complementary strategy particularly compatible with forms of regulation that seek to mitigate risks without profoundly transforming the underlying economic and technological structures.

We interpret the growing centrality of digital literacy within the European regulatory framework as a form of governance based on individual adaptation to digital risks. This approach is based on a logic in which citizens are conceived as vulnerable ‘users’ in the face of complex and potentially manipulative information environments. Therefore, the main risks arising from disinformation and AI are not addressed solely through structural interventions on platforms, but through the strengthening of individual capacities to ‘manage’ such environments. In consequence, public policies tend to assume that emerging problems in the digital ecosystem can be mitigated by improving citizens’ skills, rather than by transforming its economic and power structures. By prioritising citizens’ ability to ‘cope’ with this new environment through media and AI literacy, EU policies implicitly reinforce the idea of a public that needs to be protected through technical and educational mechanisms, due to its supposed inability to adequately process complex or misleading information (Katic 2023). For example, within the framework of the 2022 Code of Practice on Disinformation, this logic manifests itself, for example, in algorithmic content classification systems that operate without visible deliberative processes, thereby reproducing forms of technocratic governance.

The emphasis on ‘literacy’ in instruments such as the 2022 Code of Practice on Disinformation can be interpreted in line with what Morozov identifies as ‘technosolutionism’: a tendency to shift the focus from structures towards the optimisation of individual behaviour, and from political solutions to technical ones driven by technology itself (Morozov 2013). It follows that complex democratic and social problems are increasingly framed as issues that can be mitigated through technical adjustment, behavioural adaptation or improved user competences, rather than through structural transformations addressing the economic and political foundations of the digital ecosystem. Within this framework, responsibility increasingly falls on citizens, who must develop critical skills to navigate an environment designed to capture attention and extract data. As a result, technosolutionism contributes to depoliticising structural conflicts by recasting them as technical or educational challenges that can be addressed through training, resilience and digital adaptation.

In analytical terms, and following the European regulatory framework (the 2024 European Media Freedom Act and the AI Act) definitions, it is important to distinguish between media literacy and AI literacy. In the European Media Freedom Act, media literacy refers to the “skills, knowledge and understanding which allows citizens to use media effectively and safely and which are not limited to learning about tools and technologies but aim to equip citizens with the critical thinking skills required to exercise judgment, analyse complex realities and recognise the difference between opinion and fact” (European Commission 2024a, article 2, para. 21). In article 3 of the AI Act, AI literacy is defined as “skills, knowledge and understanding that allow providers, deployers and affected persons, taking into account their respective rights and obligations in the context of this Regulation, to make an informed deployment of AI systems, as well as to gain awareness about the opportunities and risks of AI and possible harm it can cause” (European Commission 2024b, Article 3, para. 56). As we will see, both operate within the same rationale of individual adaptation. This distinction is explored in greater depth in Section 5, but here it is relevant to situate how both forms of literacy form part of the same governmental logic.

Taken together, both media literacy and AI literacy function less as alternatives to technosolutionism than as some of its most visible expressions within contemporary digital governance. Both extend the regulatory logic towards the individual cognitive sphere by asking individuals to internalise risk management practices in an environment where structural conditions remain largely intact. While media literacy seeks to empower citizens to cope with the symptoms of a dysfunctional information environment, such as disinformation, manipulation or information overload, AI literacy extends this logic towards adaptation to algorithmic systems that actively mediate the production, classification and circulation of information. AI literacy seeks not only to ensure that citizens understand these systems, but also that they progressively accept the delegation and outsourcing of democratic functions to private algorithmic infrastructures.

Far from being neutral, both forms of literacy contribute to normalising a deeply asymmetrical information ecosystem, in which the responsibility for managing digital risks is progressively shifted onto users. However, the technological infrastructures that generate those risks are established as inevitable components of democratic governance. This emphasis on individual adaptation aligns with broader logics of neoliberal governance, in which the risks of disinformation and AI are individualised rather than addressed as structural political and economic problems (Drotner et al. 2017). This educational framework contributes to the depoliticisation of AI governance, and tech governance in general. By defining AI-related harms as problems of misunderstanding, misuse or insufficient competence, normative and political conflicts are reduced to matters of training and awareness-raising. In this way, power asymmetries between platforms, regulators and citizens are obscured.

Relatedly, media literacy is linked to the construction of a subjectivity (Patriarche and Zienkowski 2022) based on resilience in the face of information environments perceived as complex or potentially manipulative. This approach has been interpreted as part of a process of individualisation of risk in digital governance, in which Big Tech companies participate directly in digital literacy programmes that have been conceived as the “performance of corporate responsibility” (Docherty and Valderrama Barragán 2025). Although presented as a form of social resilience, this logic tends to reduce the problem to verification practices and information literacy, pushing its structural dimensions into the background (Frau-Meigs 2022). Indeed, the literature has highlighted a persistent tension in EU governance: despite the recognition of the systemic nature of disinformation, responses continue to be framed in terms of individual responsibility, positioning citizens simultaneously as vulnerable subjects and as the primary risk managers. Meanwhile, platforms remain the objects of incremental regulation (Casero-Ripollés et al. 2023; D’Andrea et al. 2025), although their surveillance capitalist business model remains untouched (Oleart and Rone 2025).

In consequence, business models based on data extraction and attention maximisation are treated as fixed background conditions rather than as objects of political intervention. As several studies point out, this user accountability runs parallel to limited structural intervention in platform governance, reinforcing co-regulatory arrangements that preserve platforms’ discretion whilst expanding citizens’ duties of adaptation (Pierson et al. 2023; Oleart and Bouza García 2025). Therefore, this article conceptualises the EU’s emphasis on media and AI literacy not merely as an educational policy, but as a governance mechanism. Literacy policies are therefore highly relevant: they tend to function as stabilising elements within broader governance frameworks. By emphasising individual adaptation, they contribute to normalising the centrality of platform infrastructures in democratic life, reinforcing the idea that privately owned large-scale algorithmic mediation is a structural condition of the contemporary public sphere. Consequently, literacy operates as a technology of governance that manages effects at the individual level without intervening in the underlying economic structures.

3. Methodology: Tracing Media and AI “Literacy” in EU Public Policies

The methodological approach of this article combines critical discourse analysis (Fairclough 1995; Wodak and Meyer 2015) and interpretive policy analysis (Yanow 2007), in dialogue with the literature on the critical political economy of platforms (Srnicek 2016; van Dijck et al. 2018; Zuboff 2019). The purpose of combining these approaches is to examine how EU policy documents on disinformation and AI construct problems, define responsibilities, and prioritise certain types of solutions over others. In doing so, we conceive policy documents as devices that socially construct political problems. The analysis treats policy documents not only as technical or legal texts, but as political and normative artefacts through which understandings of democracy are constructed. Attention is therefore paid both to how problems such as disinformation and AI-related risks are framed, and to how responsibility for addressing them is distributed across actors, as well as to the types of interventions that are considered legitimate within EU digital governance. The aim of the analysis is not to establish causal effects between policies and outcomes, but to identify recurring patterns in how EU institutions frame digital governance. In particular, the focus is on whether policy responses emphasise individual capacities (such as media literacy and AI literacy) or whether they also incorporate structural dimensions related to platforms, markets, and infrastructures.

The empirical corpus1 consists of EU institutional documents produced between 2018 and 2026 on disinformation and AI. The corpus is organised according to the two analytical dimensions of the study. For the analysis of media literacy, the following documents are examined: European Commission’s Action Plan against Disinformation (European Commission 2018a), the Report of the independent High-Level Expert Group on fake news and online disinformation (European Commission 2018b), the European Democracy Action Plan (European Commission 2020), the Strengthened Code of Practice on Disinformation (European Commission 2022b), the European Media Freedom Act (European Commission 2024a) and the Communication on the European Democracy Shield (European Commission 2025c). For the analysis of AI literacy, the following documents are examined: the AI Act (2024), the EU’s AI Continent Action Plan (European Commission 2025a), the Commission’s proposal for a Digital Omnibus on AI (European Commission 2025d), the Apply AI Strategy (European Commission 2025b), and the Communication on European Tech Sovereignty accompanied by an EU Open Source Strategy (European Commission 2026). Documents were selected based on their relevance to EU digital governance debates on disinformation and AI, with particular attention to instruments explicitly addressing literacy, platform regulation, or the governance of digital risks. There are some limitations to our analysis, inasmuch as we have limited our data set to the documents we conceive to be most relevant and that are illustrative of the EU’s approach to disinformation and AI.

In terms of analytical traceability, the analysis consisted of iterative and close readings of the selected documents, focusing on how responses to disinformation and AI are articulated, with particular attention to references to media literacy and AI literacy. The analytical process unfolded in two stages. In a first reading, each author independently identified and selected the textual fragments in which media literacy and AI literacy are explicitly addressed, while we maintained an overall reading of the documents in question. These fragments were then compared and, in a second reading, consolidated into the corpus of excerpts that constitute the main focus of our research. We then analysed these excerpts through interpretive reading rather than exhaustive coding and organised through a flexible analytical distinction between user-centred approaches (e.g., individual responsibility, skills, resilience, literacy) and structurally oriented approaches (e.g., platforms, infrastructures, markets, algorithmic systems). This distinction was used as a practical tool to organise the analysis rather than as a strict classification, since many policy documents combine both approaches. Overall, the methodological aim is to provide an interpretative account of how EU digital policy constructs the relationship between literacy, technology, responsibility, and democracy, and to show how this framing shapes the balance between individual and structural approaches to digital governance.

4. Results: How the EU Mobilises ‘Media Literacy’ and ‘AI Literacy’

4.1. Media Literacy and EU Approaches to Disinformation

In the EU’s institutional and political discourse, disinformation is presented as a systemic threat to democracy, electoral integrity and social cohesion. In its 2018 Action Plan Against Disinformation, the European Commission (2018a, p. 1) argues that “democratic societies depend on the ability of citizens to access a variety of verifiable information so that they can form a view on different political issues”. In the Action Plan, media literacy is already outlined as a central component of countering disinformation. Similarly, in the final report of the 2018 Commission’s High-Level Expert Group (HLEG) on fake news and online disinformation, media literacy occupies two of the five pillars: “promote media and information literacy to counter disinformation” and “develop tools for empowering users and journalists” (European Commission 2018b, p. 5). While the HLEG connects disinformation to contemporary digital infrastructures, particularly algorithmic amplification, the attention economy, and the speed and scale of content circulation on digital platforms, social media companies are perceived as part of the solution: “Online platforms are making efforts to provide responses to the distribution of disinformation” (European Commission 2018b, p. 14). Thus, although the Commission identifies systemic dynamics linked to the architecture of platforms and their business models, policy responses are largely oriented towards strategies centred on literacy, resilience and individual adaptation. For instance, it stresses that media and information literacy (MIL) is an important action line as a response to disinformation because it can empower individual users. In doing so, it will “lead to greater social resilience against disinformation and perhaps other disorders of the information age” (European Commission 2018b, p. 25).

In the 2020 European Democracy Action Plan, the Commission argues the following:

“Media literacy, including critical thinking, is an effective capacity helping citizens of all ages to navigate the news environment, identify different types of media and how they work, have a critical understanding of social networks and make informed decisions. Media literacy skills help citizens check information before sharing it, understand who is behind it, why it was distributed to them and whether it is credible. Digital literacy enables people to participate in the online environment wisely, safely and ethically.”

(European Commission 2020, p. 24)

This logic is not prominent in the first Code of Practice on Disinformation (European Commission 2022a) as there is only a single reference to media literacy, but is heavily expanded in the Strengthened Code of Practice of 2022, later adopted as a Code of Conduct within the co-regulatory framework of the DSA. In the 2022 Strengthened Code, there is a full section on “empowering users” (measures 17–25), where it is argued that “Relevant Signatories will design and implement or continue to maintain tools to improve media literacy and critical thinking, for instance by empowering users with context on the content visible on services or with guidance on how to evaluate online content” (European Commission 2022b, p. 19). This framework situates media literacy as a specific governance function, where the companies that are creating the problem in the first place are supposed to also be part of the solution by contributing to design “tools to improve media literacy”. Rather than aiming to transform the structural conditions of digital communication, it is integrated into an approach that accepts constant exposure to manipulation and information disorder as a normal condition of digital citizenship. This shift towards individualised responsibility is further consolidated through a discourse centred on critical thinking, verification and the individual assessment of content. It is in this logic that ‘fact-checking’ and networks of fact-checkers become central components of EU institutional responses to counter disinformation.

This soft approach becomes particularly evident when media literacy is articulated alongside calls for AI transparency and governance. Citizens are expected to recognise AI-generated content, interpret labels, understand algorithmic influences, and critically evaluate information flows, yet without corresponding interventions in extractive business models, attention economies, and data-driven incentives that structurally reward polarising and misleading content. Media literacy thus becomes a complementary technology of governance, aligned with AI transparency and regulation rather than a challenge to the underlying political economy of Big Tech. In fact, according to the European Commission’s Communication on “Tackling online disinformation: a European Approach” (European Commission 2018c, p. 11), AI “will be crucial for verifying, identifying and tagging disinformation”. The European Commission’s technosolutionism is also reflected in the funding of projects where AI is conceived to be able to tackle disinformation. An evident example of this perspective is the AI-Code project (https://aicode-project.eu), composed of a wide transnational consortium in which AI is mobilised to “empower media professionals” to tackle disinformation. Among other things, the project is meant to generate AI interactive coaching services addressed to media professionals, constructing AI as a technology that can meaningfully tackle disinformation (including AI-driven disinformation). This “fighting fire with fire” logic tends to downplay the business model and political economy of these companies, as Big Tech is conceived as both the problem and the solution. This logic coexists with co-regulatory frameworks such as the Digital Services Act, in which demands for platform transparency and accountability advance in a limited and negotiated manner, indirectly reinforcing reliance on literacy as a medium- and long-term solution.

This trajectory is likely to continue, as the Commission’s 2025 European Democracy Shield places particular emphasis on reinforcing the media literacy community, including strengthening the Media literacy expert group (MLEG), establishing “a new independent expert network for media literacy” and build on the work undertaken by the European Digital Media Observatory (EDMO), and “scaling-up success stories developed and deployed by EDMO regional/national hubs” (European Commission 2025c, p. 22). Altogether, the EU’s media literacy approach has been consolidated from 2018 onwards, situating media literacy as a structural policy priority of European digital governance. As Sádaba and Salaverría (2023, p. 19) point out, “recent initiatives to combat disinformation promoted by the highest European and national public institutions agree in considering media literacy as one of its main bastions”, reflecting a broad institutional consensus that “to combat disinformation and preserve democracy, it is key to train European citizens in the skillful and responsible use of information”. Furthermore, as Rone notes, most EU member states “opted for less ambitious strategies focused primarily on media literacy and educating citizens above all” (Rone 2021, p. 181), which reinforces the centrality of literacy as the predominant response within the European regulatory ecosystem.

Table 1. Media literacy in the EU policy documents on disinformation.

Document	Role of Media Literacy
Action Plan Against Disinformation (2018)	“Democratic societies depend on the ability of citizens to access a variety of verifiable information so that they can form a view on different political issues” (European Commission 2018a, p. 1)—media literacy as a central component of countering disinformation.
Report of the High-Level Expert Group on Fake News and Online Disinformation (2018)	Media literacy occupies two of the five pillars: “promote media and information literacy to counter disinformation” and “develop tools for empowering users and journalists” (European Commission 2018b, p. 5).
European Democracy Action Plan (2020)	Media literacy, including critical thinking, as an effective capacity; digital literacy enables wise, safe and ethical online participation. Citizens must check information before sharing it, understand its origin and assess its credibility (European Commission 2020, p. 24)
Strengthened Code of Practice on Disinformation (2022)	Constant exposure of users to manipulation and information disorder, framed as a normal condition of digital citizenship. Hence, there is a full section on “empowering users”. Requires signatories to design and implement tools to improve media literacy and critical thinking, providing users with context on content and guidance on how to evaluate online content.
European Media Freedom Act (2024)	Media literacy is one of the tasks attributed to the European Board for Media Services created by the EMFA. It is an independent advisory body that, among other things, is meant to “exchange experience and best practices on media literacy, including to foster the development and use of effective measures and tools to strengthen media literacy” (European Commission 2024a, Section 2, Article 13)
Communication on the European Democracy Shield (2025)	Emphasises the need to reinforce the media literacy community. Strengthen the Media Literacy Expert Group (MLEG); establishing a new independent expert network for media literacy; building on work of EDMO.

below summarises the specific ways in which media literacy is articulated in the EU policy documents we analysed.

Hence, media literacy plays a pivotal role in preventing the EU from advancing more ambitious regulatory approaches to digital platforms, such as the banning of targeted advertisement. Despite the innovative and expansive nature of the EU’s regulatory agenda, many of these instruments continue to operate within a framework that preserves the fundamental political-economic architecture of platform power (Griffin 2023; Kausche and Weiss 2025). Hence, European regulation does not necessarily challenge the structural foundations of platform capitalism, but rather tends to manage its most problematic effects through mechanisms of mitigation, oversight and social adaptation. While EU institutions recognise the systemic nature of disinformation and the structural role of platforms and algorithms, the proposed educational response ultimately reaffirms an individualised model of democratic responsibility. Citizens—often framed as ‘users’—are constructed as the primary victims of manipulation and the first line of defence against it, while platforms continue to be framed as entities in need of better regulation and supervision rather than fundamental transformation.

This logic places citizens in an ambivalent position: on the one hand, they are presented as particularly vulnerable to manipulation and disinformation; on the other, they are expected to develop the skills necessary to manage these risks themselves. In this context, media literacy functions primarily as a strategy for adapting to a digital environment already shaped by platforms, algorithms and automated amplification dynamics, rather than as a tool aimed at transforming those structures. Consequently, although EU institutions increasingly recognise the structural and algorithmic nature of disinformation, policy responses continue to focus largely on individual adaptation. Media literacy thus ends up functioning as a form of soft governance aimed at managing behaviour and strengthening citizens’ resilience within a communicative environment whose structural dynamics remain largely intact.

4.2. The Normalisation of AI Through ‘AI Literacy’

While media literacy is primarily seen as a strategy for resilience and adaptation in the face of risks structurally generated by the platform ecosystem, in the case of AI this logic takes on an additional dimension. AI literacy is presented not only as a tool for managing technological risks, but also as a mechanism aimed at facilitating the integration and normalisation of algorithmic systems across multiple spheres of social, economic and institutional life.

This role becomes more evident when AI ceases to be conceived primarily as a risk to be contained, and instead becomes a necessary infrastructural solution to problems defined as structurally unmanageable by human means. In EU’s 2024 AI Act, AI is presented as “a fast evolving family of technologies that contributes to a wide array of economic, environmental and societal benefits across the entire spectrum of industries and social activities” by “improving prediction, optimising operations and resource allocation, and personalising digital solutions” (recital 4). In fact, AI “should serve as a tool for people, with the ultimate aim of increasing human well-being” (recital 6). In relation to literacy, recital 20 states that “in order to obtain the greatest benefits from AI systems while protecting fundamental rights, health and safety and to enable democratic control, AI literacy should equip providers, deployers and affected persons with the necessary notions to make informed decisions regarding AI systems”. AI is therefore presented as a technology that will have many benefits, including in education. For instance, the AI Act also suggests the following: “The deployment of AI systems in education is important to promote high-quality digital education and training and to allow all learners and teachers to acquire and share the necessary digital skills and competences, including media literacy, and critical thinking, to take an active part in the economy, society, and in democratic processes” (recital 56 of the AI Act).

The technosolutionism embedded in the EU’s AI Act has only increased in 2025 during the second Commission presidency mandate of Ursula von der Leyen. In the EU’s 2025 AI Continent Action Plan, AI literacy is situated as a central pillar: “Developing a broad-based AI-savvy workforce starts with high-quality and inclusive initial education and training. The 2030 Roadmap on the future of digital education and skills and its AI in Education initiative, will support the development of AI literacy for primary and secondary education and foster the strategic and ethical uptake of AI in education, including through support and capacity building for teachers and education institutions” (European Commission 2025a, p. 18). Coherently, in parallel to this Action Plan the Commission put forward the “Apply AI Strategy”, where one of the goals is to promote an “AI-ready workforce across sectors”, and “provide access to practical AI literacy trainings tailored to sectors and job profiles through the AI Skills Academy” and it is established that “to ensure a responsible and beneficial use of AI among all workers, adequate skills are a prerequisite. Solid AI literacy should start at an early educational level and continue to the labour market through reskilling and upskilling” (European Commission 2025b, p. 14).

Furthermore, the recent 2025 Digital Omnibus package on AI presented by the European Commission only exacerbates the problem. Under the AI Act, providers and those responsible for deploying AI technology (also known as Big Tech companies) were required to ensure AI literacy. However, amongst other things, the Commission’s initial proposal for the Digital Omnibus Package removes this responsibility and, instead, suggests that the AI Act ‘should be amended to require Member States and the Commission, without prejudice to their respective competences, to encourage, individually, collectively and in cooperation with relevant stakeholders, providers and those responsible for deployment to provide a sufficient level of AI literacy to their staff and to other persons involved in the operation and use of AI systems on their behalf’ (European Commission 2025d, pp. 12–13). Consequently, Big Tech companies would no longer be obliged to ensure AI literacy, and the Commission and Member States would take the additional responsibility for AI literacy. To be sure, the Digital Omnibus on AI we analysed is the initial proposal, and the final version will likely maintain the AI Act obligations of providers and deployers to ensure AI literacy, primarily because EU member states are reluctant to take on additional commitments in this field. However, the fact that the Commission even proposed this measure is illustrative of its strong push to encourage public authorities to normalise AI across society.

The normalisation and dependence on AI—and on the small group of companies capable of developing and scaling it—may contribute to reconfiguring the balance of power in ways that favour Big Tech and raise concerns about implications for democratic governance. Once literacy becomes the political horizon, the expansion of AI is implicitly treated as inevitable, and the burden of adjustment is shifted onto individuals, institutions and workers. From this perspective, the problem is not education itself, but the political horizon implicit in the call for ‘AI literacy’. Why must the default democratic goal be for citizens to learn to use and integrate AI into their daily lives, rather than collectively debating whether, where and under what conditions AI should be deployed? In this configuration, AI literacy becomes a vehicle for normalisation: it can present the spread of AI as something desirable and inevitable, and may diminish the significance of democratic agency. Citizens are positioned as competent users and adaptable subjects, rather than as political actors capable of questioning the infrastructural decisions that reshape the public sphere. AI literacy is even included in the June 2026 “Communication on European Tech Sovereignty, accompanied by an EU Open Source Strategy”, where the Commission put forward the “European Technological Sovereignty Package”. In an increasingly geopolitical framing of AI (where the EU is perceived to be lagging behind the United States and China), the European Commission (2026, p. 15) argues that in “order to achieve technological sovereignty, education systems must build competences cumulatively—from basic skills in compulsory schooling through higher and vocational education to lifelong learning. Foundational digital competences and AI literacy acquired in schools are the bedrock on which specialised expertise is later built”.

Table 2. AI literacy in EU policy documents on artificial intelligence.

Document	Role of AI Literacy
AI Act (2024)	“AI literacy should equip providers, deployers and affected persons with the necessary notions to make informed decisions regarding AI systems” (recital 20), in order to protect fundamental rights and enable democratic control.
AI Continent Action Plan (2025)	Develop AI literacy for primary and secondary education, fostering uptake of AI in educational institutions.
Digital Omnibus on AI (2025)	The Commission proposed to shift the responsibility of AI literacy to Member States and the Commission instead of providers and deployers.
Apply AI Strategy (2025)	Push to promote AI literacy from “early educational level and continue to the labour market through reskilling and upskilling” (European Commission 2025b, p. 14)
Communication on European Tech Sovereignty/EU Open Source Strategy (2026)	“Foundational digital competences and AI literacy acquired in schools are the bedrock on which specialised expertise is later built” (European Commission 2026, p. 15)

below summarises the ways in which AI literacy is discussed in the EU policy documents we analysed.

All in all, AI literacy plays a central role in the institutionalisation of algorithmic systems within digital governance. By normalising AI across society, the EU is facilitating adaptation to increasingly automated forms of decision-making (Laux et al. 2023; Porlezza 2023). AI literacy can then be understood as serving not only an educational function but also one of Big Tech legitimisation, making the growing dependence on privately owned automated systems more socially acceptable. We interpret this dynamic as fostering technosolutionist logics, where trust in algorithmic systems partially replaces public deliberation. Rather than reducing dependence on algorithmic governance, AI literacy contributes to its normalisation. Rather than questioning the design of these systems, it focuses on empowering individuals to interact with them more efficiently. This shifts the notion of digital citizenship towards forms of technical competence rather than towards democratic participation in the design of the infrastructures that structure the public sphere.

Taken together, the findings reveal a consistent pattern across the two domains under analysis. In both media literacy and AI literacy, the proposed solutions tend to be oriented towards building competences for individuals. Examples include training programmes (Apply AI Strategy, AI Continent Plan, Digital Omnibus on AI), user empowerment tools (Strengthened Code of Practice on Disinformation, Report of the High-Level Expert Group on Fake News and Online Disinformation), or calls for citizens to verify information before sharing it (European Democracy Action Plan). Illustratively, the European Tech Sovereignty package explicitly frames technological sovereignty as partially contingent on the development of AI literacy as a necessary condition for individuals to build their own future expertise. At the same time, while some documents acknowledge the structural role of Big Tech companies in fostering disinformation and bad usages of AI, no emphasis is placed on the business models of Big Tech companies, built on surveillance, virality, user engagement, the use of algorithms for targeted advertising (Diaz Ruiz 2025), and more generally surveillance capitalism (Zuboff 2019). This is why we return to the political economy of Big Tech, comparing media literacy and AI literacy as governance frameworks, and suggest an alternative conception of literacy better suited to address the structural dimensions of the problem.

5. Discussion: Literacy, Technosolutionism and the Political Economy of Big Tech

Our empirical work has shown how literacy approaches aimed at addressing the risks associated with contemporary digital technologies tend to be based on a predominantly individual-oriented approach. Literacy approaches are likely to prove insufficient to address the harms arising from digital platforms and AI systems, and may even be associated with reinforcing the very dynamics they seek to mitigate, by shifting the focus towards solutions centred on user behaviour rather than on the underlying structures (Oleart and Rone 2025). This tension manifests itself differently in the two main approaches promoted within the EU framework: media literacy and AI literacy.

As we have previously shown in Section 4, the ‘literacy’ approach as a solution to disinformation or AI-related challenges shifts the focus of intervention from structures and infrastructure towards individual citizens or ‘users’. However, there are significant differences between media literacy and AI literacy. Media literacy remains normatively desirable in democratic societies, although, as currently mobilised by policy-makers, it is often reduced to individualised resilience and fact-checking practices. By contrast, ‘AI literacy’ is a politically ambivalent concept. It can refer both to a critical understanding of AI systems (their limits, biases, energy consumption, material infrastructure and power relations), and to a more instrumental competence aimed at accelerating their adoption and integration into everyday life. In EU policy practices, AI literacy typically functions as the latter: a framework for normalising the spread of AI in social, professional and political spaces, thereby potentially reinforcing dependence on the corporate actors who control its development and deployment.

The patterns of intensive use associated with social media and AI-based technologies cannot be understood solely as individual choices, but also as the result of digital infrastructures designed to maximise user attention and engagement. Various studies have shown that the interfaces and algorithms of digital platforms are deliberately configured to encourage repetitive and potentially compulsive usage patterns, integrating these addictive behaviours into business models based on data capture and targeted advertising (Srnicek 2016; Zuboff 2019; Wiesböck et al. 2026). Indeed, Big Tech companies—and particularly those focused on social media and AI—profit immensely from targeted advertising that is only possible due to their surveillance capitalist business models. Reportedly, Google, Amazon and Meta are projected to account for as much as 62.3% of 2026 digital ad spending globally (Singh 2026). In this context, the longer users interact with these platforms, the greater the amount of data generated and the economic value that companies can extract from it, reinforcing an attention economy in which communication, information and social interactions are increasingly mediated by digital infrastructures largely controlled by a small number of large technology companies.

What media literacy and AI literacy have in common is that, as currently promoted within EU frameworks and regulations, they tend to shift the focus of the debate away from the political economy of platforms and towards the individual capabilities of users. In other words, the emphasis on ‘literacy’ helps to obscure a fundamental structural issue: who owns and controls the infrastructures through which democratic publics communicate, deliberate and form their political preferences. As van Dijck points out, the central questions are not merely what content circulates, but ‘who owns and exploits the data flows, who controls algorithmic governance, and who is responsible and accountable for its impact’ (van Dijck 2020, p. 3). This omission is particularly relevant given that the privatisation of the public sphere has historically been recognised as a persistent democratic problem. Liberal democracies have always had a tension with big private communication infrastructures—from the print media to television—and the concentration of ownership in the hands of figures such as Rupert Murdoch or Silvio Berlusconi has shown how media pluralism, the setting of the public agenda and even electoral competition can be structurally conditioned by private power.

This problem does not disappear with the emergence of social media and AI, but rather intensifies in the age of tech platforms. Contemporary tech oligarchs such as Elon Musk or Mark Zuckerberg do not represent a break with the old media magnates, but rather their extended continuity within a different infrastructural context. Unlike traditional media owners, their influence is not exercised solely through editorial lines or editorial decisions, but through direct control of the infrastructures of visibility: ranking, recommendation and amplification systems that determine what circulates, what is seen and what remains invisible in the digital public sphere. These mechanisms are largely opaque and escape the control of both users and governments. The result is a form of power that is not merely ideological, but infrastructural and systemic, exercised through technical decisions that structure the conditions of public deliberation in advance. In this sense, the power of contemporary platforms is part of a broader democratic problem: communication infrastructures increasingly controlled by corporations.

This dynamic becomes apparent in specific cases that illustrate how infrastructural control translates into reconfigurations of the public sphere. Elon Musk’s 2022 acquisition of Twitter (rebranded as X in 2023) demonstrates how changes to a platform’s moderation and governance rules can rapidly alter the conditions of visibility for public discourse, framed discursively in defence of ‘freedom of expression’, although with uneven and often opaque effects on algorithmic visibility systems. Similarly, Jeff Bezos’s ownership of The Washington Post highlights how even traditional media institutions can be progressively integrated into broader corporate logics, in which the editorial function coexists with structural conflicts stemming from billionaire ownership and interdependence with economic and regulatory interests. In both cases, what emerges is a hybrid figure of power, in which figures such as Bezos, Zuckerberg or Musk—who are also deeply involved in the AI ecosystem—can be seen as contemporary versions of old media oligarchs such as William Randolph Hearst (depicted as “Citizen Kane” in the famous 1941 film directed by Orson Welles). What they share is not only the ability to influence public opinion, but control over the very infrastructures that make that opinion possible.

Hence, the underlying issue cannot be reduced to improving media or digital literacy, but there must be a focus on the ownership structure and economic organisation of the digital ecosystem. The infrastructures of democratic communication—including traditional media, digital platforms and even technical components such as undersea cables or satellite systems like Starlink—tend to be treated as private assets rather than public goods. From this perspective, strategies focused exclusively on literacy prove insufficient, insofar as they do not address the dynamics of power concentration, profit incentives and the business model of platform capitalism, based on the maximisation of attention, the prediction of behaviour, targeted ads and the circulation of polarising or sensationalist content. As long as this business model remains based on the extraction of attention and data on a large scale, any approach based solely on media literacy will operate at a structural disadvantage: citizens are asked to develop self-regulatory capacities within environments designed precisely to erode them. Literacy ceases to be a neutral response and comes to function as a form of governance that in practice contributes to stabilising oligarchic control over the infrastructures of democratic life.

Finally, the implications of promoting ‘AI literacy’ when it is conceived primarily as a tool for its adoption and social normalisation are not only epistemic and social, but also material and ecological. Even if AI is increasingly presented as a solution to social problems (including climate change), the widespread normalisation of AI relies on energy-intensive infrastructure (in particular, the growing number of data centres), global supply chains and data extraction practices. Recent critical reports have highlighted how the contemporary AI boom is inseparable from energy-intensive infrastructure, which complicates narratives presenting AI as a simple climate solution (Hao 2025). Indeed, researchers have shown how Big Tech’s AI Greenwashing efforts might be a fruitful avenue for climate litigation under EU consumer protection laws, which in turn might attract political attention and politicise the question of whether AI should be normalised (Griffin and Sander 2026). Relatedly, there is evidence of labour exploitation towards communities located primarily in the Global South (Cant et al. 2024) and an extractivist logic that requires increased extraction of critical raw materials (Scheyder 2024). The current political emphasis on ‘AI literacy’ that we have empirically traced contributes to rendering these material, labour and ecological dimensions invisible. AI literacy is thus reduced to a form of functional adaptation to existing infrastructures, rather than opening up a space for their critique. Hence, a conceptual and normative contribution of this article is to broaden the notion of AI literacy towards a more critical and structural understanding of the socio-technical systems within which it is embedded. This implies not limiting it to a technical or adaptive competence, but understanding it as the capacity to situate AI within broader data infrastructures, economic incentives and power relations, including its ecological and labour dimensions.

6. Conclusions: Reclaiming Democratic Governance and Ownership of the Technologies That Shape Our Societies

We began our article by asking what a ‘literacy’ approach means for policymakers’ efforts to democratise technology. After examining the specific ways in which “literacy” is deployed in EU public policies, we argue that the emphasis placed by EU policy-makers on both ‘media literacy’ and ‘AI literacy’ is misguided. This is particularly the case when it comes at the expense of more structural approaches to tackling the rise of disinformation and AI. We argue that EU policymakers’ emphasis on literacy places the responsibility on individuals rather than on the system in which they operate. As a result, literacy becomes a strategy of adaptation rather than transformation, making it even more difficult to tackle these structural problems.

In the case of media literacy and disinformation, the central political problem is not disinformation alone, but also the highly concentrated and monopolistic market structure of both social media and traditional media, as well as the business models under which they operate. As we argue throughout this article, this concentration of infrastructural and algorithmic power may also enable these actors to mediate which issues and narratives become more visible in the public sphere, and which remain comparatively marginalised. A similar logic applies to AI, albeit with potentially more far-reaching implications. As we have argued, AI literacy frameworks also risk normalising the expansion of AI as socially inevitable or inherently desirable. For example, AI technosolutionism is increasingly visible in the realm of citizen participation, where AI tools are promoted as ways to ‘improve’ deliberation, while potentially undermining democratic debate itself (see Oleart and Palomo 2025). This dynamic resonates with the administrative and depoliticised conception of democracy that characterises the EU—and the Commission in particular (Oleart and Theuns 2023)—which helps to align with technosolutionist interpretations of political decision-making and reinforces the idea that the misuse of AI is primarily the responsibility of users rather than the companies that develop and deploy these systems. We are of course not against the idea of literacy or education in the digital field of media and AI. However, the particular policy usages of literacy that we have uncovered are problematic.

In both cases, literacy-led policy responses risk prioritising individual adaptation over structural transformation, leaving underlying power concentrations largely unaddressed. It is necessary to broaden the debate beyond individual adaptation and consider alternatives aimed at transforming the structural conditions upon which digital technologies are developed. If the aim is to democratise the digital ecosystem, the debate must go beyond individual adaptation and confront and democratise the political economy of the platforms that organise contemporary communication. Current debates on the expansion of AI raise questions about its social, environmental and labour implications, as well as the growing concentration of power in the hands of a small number of Big Tech companies and tech oligarchs. Recognising these tensions requires moving beyond narratives of inevitable technological progress and reopen the democratic debate on how technologies should be designed, governed and deployed. In this regard, historical experiences of technological change also suggest that such processes have consistently been shaped by political conflict rather than technological inevitability. The Luddite movement in early nineteenth-century England, for instance, did not constitute a rejection of technology as such, but rather expressed struggles over the distribution of power and benefits associated with mechanisation in the textile industry, in a context of rapidly changing labour relations (Merchant 2023). Technological change has always been embedded in disputes over governance, power relations and distributional outcomes, rather than unfolding as a neutral or predetermined process. Therefore, a more ambitious conception of media, digital and AI literacy should not be reduced to individual adaptation to the existing technological environment, but should foster a critical engagement with the socio-technical systems, including the epistemological, ethical and relational conditions of AI, as well data infrastructures, economic incentives and power relations that shape the contemporary digital ecosystem (Rapanta et al. 2025; Robinson and Hollett 2024).

Ultimately, the central argument of this article is that systemic problems require systemic responses. While literacy initiatives may help citizens navigate complex information environments, they cannot replace policy interventions that address the concentration of power, the surveillance capitalism business model of Big Tech and their structural control over current communication infrastructures. Reclaiming democracy in the digital age therefore requires not only more informed users, but also more democratic governance and ownership of the technologies that shape our societies.