Multifocal analysis of narratives and representations of the European AI Act in Spanish digital newspapers

Online Journal of Communication and Media Technologies, 2026 , 16 ( 3 ), e202643 e - ISSN: 1986 - 3497 Copyright © 2026 by authors; licensee OJCMT by Bastas , CY . This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution License ( https://creativecommons.org/licenses/by/4.0/ ) . OPEN ACCESS Multifocal analysis of narratives and representations of the European AI Act in Spanish digital newspapers María Dolores Meneses - Fernández 1* 0000 - 0001 - 8516 - 4298 María Sonnia Chinea Rodríguez 1 0000 - 0002 - 5811 - 4882 1 Department of Communication Sciences and Social Work , Faculty of Social and Communication Sciences , Universidad de La Laguna , Canary Islands, S PAIN * Corresponding author: dmeneses@ull.edu.es Citation: Meneses - Fernández , M. D ., & Chinea Rodríguez , M.S . ( 2026 ). Multifocal analysis of narratives and representations of the European AI Act in Spanish digital newspapers . Online Journal of Communication and Media Technologies, 16 ( 3 ), Article e202643 . https://doi.org/10.30935 /ojcmt/ 18968 ARTICLE INFO ABSTRACT Received: 18 Jul 2025 Accepted: 20 Jun 2026 This study analyzes Spanish news coverage of the AI Act from a European perspective, still unexplored in specialist literature. We examine how journalistic narratives are constructed surrounding the risks associated with AI, digital sovereignty and the tension between security - rights and technological innovation. We applied a multifocal theoretical framework and mixed method research design with an inductive approach to the coding, categorization and news analysis of a textual corpus derived from Spanish digital newspapers with high readership rates, published between 2020 and 2024. The results reveal narrative imbalances in the representation of pro - digital rights NGOs and Big Tech stakeholders concerned by European Union regulatory processes. Moreover, the analysis shows that journalistic treatment of these complex regulations and AI technology has been undertaken with a digital literacy deficit. These narratives seem more directed at politicians, experts and interested organizations than to end users, even though they are the ones the AI Act is designed to protect . Keywords: journalism, AI Act, AI regulation, artificial intelligence, media analysis, European Union INTRODUCTION : BACKGROUND AND PURPOSE Faced with the expansion of a rtificial i ntelligence (AI) and the need for risk management, the European Union (EU) initiated a regulatory process in 2021 to establish a legal framework for AI development and use. The European Commission (EC) was advised by the High - level Expert Group on AI (AI - HLEG), which put forward both ethical recommendations (AI - HLEG, 2019a) and policy recommendations (AI - HLEG, 2019b). In addition, the EC also launched a public consultation. This consultation revealed that civil society favors the use of strict rules, while the technology industry and tech companies would prefer an absence of excessive regulations so as not to hinder innovation (EC, 2020). These results influenced the drafting of the AI Act proposed by the EC (2021). The process culminated on March 13, 2024, when the European Parliament passed Regulation (EU) 2024/1689 (AI Act) to define the regulatory legal framework for AI systems, particularly those deemed to be high - risk (Pehlivan, 2024; Wagner et al., 2024). The AI Act came into force on August 1, 2024, but will not come into full effect until after August 2, 2026. The EU has adopted a risk - based approach (RBA) aimed at ensuring that AI systems marketed in the EU are safe and respect fundamental rights (Pehlivan, 2024). The AI Act explicitly bans certain practices, establishes requirements for high - risk systems, and sets forth strict requirements on transparency obligations & AI practices. To monitor compliance, the AI Act proposes: Research Article Meneses - Fernández & Chinea Rodríguez 2 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 (1) the creation of a European Artificial Intelligence Council to oversee and enforce compliance across member states and (2) the introduction of market surveillance mechanisms. Academics agree that the AI Act is likely to have a global impact, becoming a model for governance in much the same way as the g eneral d ata p rotection r egulation (GDPR) did (Gasser, 2023 ; Pehlivan, 2024 ). There can be no doubt that the AI Act is landmark legislation that is deserving of examination from the perspective of news media as well. The creation of this legislation has captured the attention of the Spanish news media, whose coverage is vital given that it conveys not only the legal ramifications of the Act, but also an understanding of the technology being regulated. Our study analyzes how the underlying negotiations and final adoption of the AI Act were reported by Spanish news media between 2020 - 2024. In doing so, we have identified the narratives, representations, and frames disseminated by Spanish newspapers as information gatekeepers and the architects of public agenda - setting (McCombs & Valenzuela, 2021; Weaver et al., 2004). From the US A perspective, Hine (2024) recognizes that the AI Act sets global precedents. The expansion of ultra - European legislation has therefore solidified interest in unearthing the dominant narratives behind the social construction of digital governance, identifying asymmetries between technical and media discourses, and uncovering representations of tensions or dilemmas. As previously mentioned, globally speaking the AI Act is a piece of landmark legislation that merits analysis from the perspective of news media. With this in mind, we examined four private Spanish media outlets. This analysis provides insights on deliberative pluralism and also on the discursive structures employed in the sources selected by journalists (van Dijk, 1990). For this reason, our interest in identifying the journalistic narrative and voices seeking to influence the AI Act, whether it be through techno - transnational strategies or social strategies. LITERATURE REVIEW Given that the AI Act is so recent, there are currently no systematic studies on media coverage of AI from a European perspective nor of Spanish digital newspapers; consequently, neither the narratives being constructed (broadcast) nor their effects on the public (reception and perception) are known. To date, the studies published focus on AI regulation and only make a passing reference to the media. As far as we have been able to ascertain, the studies that have been published (WoS and Scopus) on the AI Act from a Social Sciences perspective focus on legal and technical aspects, ethical and legal issues, politics, foreign affairs and diplomacy, business, or RBA and security. Therefore, by examining the Spanish media ’ s coverage of the AI Act, which is landmark legislation, rather than coverage of AI in general, this study provides original insights and contributes to the literature. Using the EU as an example, Horowitz et al. (2018) stress that the USA needs a national AI strategy, with a focus on regulation and governance to ensure security safeguards are in place and to offset any imbalance in global power. They propose that a government initiative of this kind should be comprehensive and, thus, also consider interaction with the media and information space. In Ouchchy et al. (2020), three contributions on AI regulation in the USA stand out: the first is the mismatch between the ethical issues prevalent within academic literature and those that actually receive the most media attention (unemployment, privacy, security, and human control); the second is the implications of this mismatch for the public ’ s understanding of urgent regulatory initiatives in the US A and non - EU countries; and the third is the characterization of current coverage of AI as accurate, but superficial, and lacking a multifaceted approach required to provide adequate information. The study by Brennen et al. (2021) on AI reporting within the British media identifies a cyclical pattern of utopian - dystopian narratives in the coverage of regulatory activities. It shows that coverage of regulatory initiatives responds more to the dynamics of expectations than to objective technical assessments; it also shows that journalistic frames are more influenced by sociocultural factors than by technical or legal considerations. Tolmeijern et al. (2021) and d e Almeida et al. (2021) identify the media ’ s tendency to simplify the ethical principles underlying AI regulatory proposals. For these authors, this oversimplification creates black - and - Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 3 / 22 white false dichotomous representations of multifaceted & complex ethical debates, thereby affecting the public ’ s understanding of complex regulatory initiatives. Allaham et al. (2025) studied media from 27 countries. They identified that AI coverage prioritizes social, legal and human rights risks, i.e., those affecting a society ’ s social structure and its legal system. They observed ideological patterns and found: the right - wing media accentuates AI ’ s existential risks; the left - wing European media accentuates risk of harm to society or to people ’ s legal rights; the North American media accentuates cognitive risks; and centrist media demonstrates less bias and polarization. The few studies that address media coverage in relation to AI governance do not focus specifically on the AI Act in the Spanish context, but rather on broader representations of AI and its regulation in European media ecosystems (Gourlet et al., 2024; Helberger, 2024 ; Porlezza, 2023, 202 4 ). In view of this deficient landscape in journalistic discourse analysis, it is understandable that Helberger and Diakopoulos (2023) call on the academic community to better understand which uses of AI put democratic values at risk, and how AI can contribute to the functioning of the media in a democracy. THEORETICAL FRAMEWORK In this study, we use a multifocal framework to develop an integrated theoretical and conceptual approach involving various levels of analysis (discourse, representation, narrative, and framing) ( Figure 1 ). We draw on the existing links between these perspectives (Burscher et al., 2016; Ørmen & Gregersen, 2019) as well as between these and risk studies (Beck, 1998; Kasperson, 1992; Kasperson et al., 1988, 2022). We believe that news reports on the AI Act narrate, represent, and frame the regulation of a technology associated with unexampled risks ( Figure 1 ). We analyze the risks of AI in news coverage of the AI Act from a perspective that integrates risk theory (Beck, 1998; Kasperson, 1992), agenda - setting (McCombs & Valenzuela, 2021; Weaver et al., 2004) and framing analysis (D ’ Angelo, 2019; Entman, 1993, 2007). This combination allows us to examine how the media construct narratives about technological risks when covering technological regulation. We explore how journalists and the media address these risks during a regulatory process based on the RBA , seeking to determine which types of risk are emphasized and how they are portrayed in relation to uncertainties, legal loopholes, and challenges in terms of enforcement of the AI Act. Figure 1 . Diagram of integrated theoretical - conceptual framework & study methodology (the a uthors ’ own elaboration) Meneses - Fernández & Chinea Rodríguez 4 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 Complementary T heoretical F rameworks Risk theory (Beck, 1998; Kasperson, 1992) provides symbolic and communicative dimensions to the critical understanding of technologies and their regulation. Beck (1998) postulates a transition towards the risk - based society, characterized by transnational manufactured risks that can potentially impact humanity as a whole. These concepts are inherent to AI, a frontierless technology unrestrained by physical borders that poses real challenges in terms of global governance. According to Kasperson (1992 ) and Kasperson et al. ( 1988, 2022), the media hold considerable sway over public perceptions of technologies and their risks and actively participate in their social construction. Agenda - setting (McCombs & Valenzuela, 2021; Weaver et al., 2004) helps to uncover the points being highlighted, the thematic hierarchies, and the approaches that receive the most attention in news coverage of the AI Act, shaping the narratives influencing public perception. Framing analysis (D ’ Angelo, 2019 ; Entman, 1993, 2007 ) reveals the aspects emphasized by journalists in accordance with their ideological and perceptual frames. Journalists, in acting as ideological agents and sense - makers (Koziner, 2013), structure how issues are presented in a way that influences their audience ’ s attitudes (Scherrer, 2023). This approach examines how discourses surrounding the AI Act are constructed, which aspects are emphasized, and which stakeholders receive more attention; and, in doing so, it reveals the tensions that arise during the regulatory process. Operating C oncepts Technological risk (Beck, 1998; Kasperson, 1992; Kasperson et al., 1988, 2022) is a central pillar for understanding social responses to AI. As a disruptive technology, AI creates uncertainties that the media can either heighten, play down, or reframe. An analysis of the media narratives surrounding these risks reveals both shared concerns and journalistic constructs that shape public understanding. Digital sovereignty emerges due to its connection to the risks that the AI Act seeks to address (Pohle & Thiel, 2020). However, Couture and Toupin (2019) and Falkner et al. (2024) caution that it is not a monosemic concept. The literature reflects this diversity: Dammann and Glasze (2023), Falkner et al. (2024), Floridi (2020) and Pohle and Thiel (2020) speak of digital sovereignty ; Grant (1983), and also Dammann and Glasze (2023), Falkner et al. (2024) and Walker (2024) speak of technological sovereignty ; Lane (2022) introduces the term data sovereignty ; while others allude to self - determination , informational sovereignty or digital autonomy . The World International Forum (Fleming, 2025) defines it as “ control over your own digital destiny ” . In the EU, it implies the autonomy to exercise “ control of the digital on the physical layer, the code layer, and the information layer ” (Falkner et al., 2024, p. 2102). Floridi (2020, p. 377) warns that “ whoever will win the fight for digital sovereignty will determine the lives of all people on both sides of the digital divide ” . Media narratives provide a framework for analyzing news articles as texts subject to journalistic conventions (Buozis & Creech, 2017). This approach examines how stories are told (Hayles, 1999; Zylinska, 2020) in specific contexts. Buozis and Creech (2017, p. 1431,1433) define narratives as “ a way of organizing the world ” that participates in “ the construction, maintenance, and dissemination of cultural myths ” , allowing us to analyze how the regulatory process of the AI Act is narrated. Tone (or sentiment analysis) is a critical dimension of journalistic communication. It is described as the subjective slant or evaluative bias that the media imbue in their content (Hopmann et al . , 2010; Jacobs, 2020; McCombs & Valenzuela, 2021; Pollach & Hansen, 2021; Sparks & Hmielowski, 2022; Weaver et al., 2004). Tone acts as an evaluation mechanism that transcends the mere presentation of facts and serves to influence public opinion on issues such as politics, the economy, or equally, social issues & affairs. As an integral part of the second level of agenda - setting and framing theory, it assigns values to the issues under the spotlight. This complementarity demonstrates how the media chooses which aspects of reality to cover when reporting (framing) and that it assigns them an evaluative burden (tone) that shapes public perception. Tone categorization is usually limited to positive , negative , or neutral (Sparks & Hmielowski, 2022), but we have expanded the categorization to include others adapted to the coverage of the AI Act. Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 5 / 22 Actors and P ower D ynamics The concepts of digital activism and corporate lobbying ( Appendix A ) have also been applied, as they represent the opposing forces vying to influence the drafting process of the AI Act (Bennett & Segerberg, 2012; Hine, 2024; Sharon & Gellert, 2024). The analysis of agenda - setting and framing allows us to identify the representation of competing interests from within media coverage, thereby revealing underlying power dynamics and the media ’ s (in)visibilization of different actors. The combination of these approaches provides an analytical framework for identifying meaning construction in journalism surrounding the AI Act. This allows us to understand the narratives that shape public dissemination surrounding a regulation that will undoubtedly have an impact beyond the EU ’ s borders, and likely shape international legislation (Feldstein, 2024; Hoffmeister, 2024; Pagallo, 2023). Research Q uestions and O bjective Given the informational context surrounding the AI Act, and with the objective of identifying the topics and perspectives emphasized by the media, we formulated the following four research questions (RQ s ): RQ1 : On alluded risks: What categories of risks do journalists/the media emphasize? RQ2 : On EU digital sovereignty: In the newspapers, what connections emerge between public consultations & negotiations and approval of the AI Act, and the narratives surrounding the EU ’ s strategic autonomy? RQ3 : On digital activism vs. corporate lobbying: How do journalistic narratives represent the conflict of interests between digital activism and corporate lobbying when reporting on the AI Act? RQ4 : On journalistic tone (sentiment analysis): Which journalistic tones are associated with the Spanish media ’ s coverage of the AI Act? MATERIALS AND METHODS We applied qualitative analysis of textual data. Corpus and N ewspapers A nalyzed We selected four widely - circulated Spanish digital newspapers with large readerships, based on readership rankings for unique users and average audience (GfK DAM and Scimago Media Rankings). Newspaper names: El Español , El País, El Mundo, and El Diario . The chronological timeframe selected (January 2020 - August 2024) spanned from the launch of the petition for AI regulations through to the passing of Act and resulting reactions. Selection and C ollection C riteria We used exhaustive sampling, which consisted of selecting all the published pieces that met the established temporal and thematic criteria, locating the texts using search engines, and then locating the following terms: “ Artificial Intelligence Act ” , “ AI Act ” , “ European Regulation on Artificial Intelligence ” , “ REAI ” , “ Artificial Intelligence Act ” and “ AI Act ” . The corpus consisted of 306 items, which were manually filtered to remove elements that could not be analyzed. The elements removed due to being unanalyzable included duplicated items, incomplete texts and pieces consisting solely of a headline or one - sentence newsbite lacking any narrative development related to the AI Act; they also included items limited to multimedia content without substantive textual information. Category S ystem D evelopment Category ‘ t hematic approach ’ We based this category on the concept of thematization (Saperas, 1987), which is interlinked with agenda - setting and framing. We grouped themes by conceptual connections or related meanings until we finalized the codes comprising this category (Strauss & Corbin, 2016). Meneses - Fernández & Chinea Rodríguez 6 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 Category ‘ j ournalistic t one ’ We analyzed the evaluative orientation of journalistic treatment following the methodology established in the literature ( Hopmann et al . , 2010 ; Jacobs, 2020; Pollach & Hansen, 2021), which defines tone as an evaluative component capable of influencing audiences. Some authors equate this with sentiment analysis (Sparks & Hmielowski, 2022). For greater precision, we expanded the standard ratings of negative , neutral , and positive to include others related to the subject under analysis. Risks : We specify the type of risk, the actors, and the positions taken. Digital sovereignty : We establish this through allusions to the EU ’ s role in pioneering regulations. Digital activism vs. corporate lobbying : We establish this through allusions to civil organizations and large corporations that have a vested interest in AI and in influencing the drafting of the AI Act. Analytical T echniques Our coding strategy followed a predominantly inductive logic, in line with g rounded t heory principles (Strauss & Corbin, 2016), while being informed by risk theory, agenda - setting and framing as sensitizing concepts. We first allowed themes and tones to emerge from the corpus through open coding and constant comparison, and only then related them to the theoretical dimensions identified a priori. This abductive combination of inductive emergence and deductive theoretical alignment enabled us to iteratively refine the categorical system and to connect empirical patterns with the multifocal theoretical framework. We applied semantic network analysis based on risk theory, agenda - setting, and framing, as outlined in the theoretical - conceptual framework. Inductive qualitative analysis was combined with different quantitative parameters: groundedness 1 , defined as a quantitative indicator for measuring the frequency with which each code (thematic approach or tone) was assigned to citations within the corpus, capturing its depth and relevance; co - occurrences 2 , which measure the association between thematic approaches and/or tones (i.e., the frequency with which two codes appear together in the same passage, indicating interdependencies); and density 3 , understood as the number of direct links (co - occurrences) each code maintains with others in the network, which we interpret as an indicator of connectivity and conceptual complexity. This operationalization of groundedness, co - occurrences, and density follows the tradition of g rounded t heory coding (Strauss & Corbin, 2016) and semantic network approaches for media texts (Burscher et al., 2016). We opted to use ATLAS.ti v23 as analytical software, which facilitated automated initial coding and served as a host environment for the content analysis, providing natural language processing (NLP) ‑ assisted suggestions for developing the final codebook and generating semantic network visualizations. We carried out these steps with constant manual verification in an iterative process (Burscher et al., 2016; Nguyen & Hekman, 2022). Iterative C oding P rocess We started with approximate and systematic coding using analytical software. We combined NLP from the initial automated coding with human coding to verify that each coded quotation reflected both the thematic approaches relating to the AI Act, and the journalistic tone used by the reporter. We scaled the tones progressively through NLP and manual verification (Pollach & Hansen, 2021). Coding was finalized upon reaching sample saturation. 1 Groundedness represents the empirical strength of a category or code. It is the total number of quotations or text segments in the corpus that are associated with a specific code. High groundedness indicates that the category or code is robust in the corpus and functions as an indicator of frequency and relevance in the documents analyzed. 2 Co - occurrence measures the intersection of two or more codes within the same text segment. It identifies emerging relationships and patterns of association between concepts or ideas, and its analysis helps to determine whether the joint appearance of two categories or codes is statistically meaningful or merely due to chance. 3 Density reflects the theoretical strength or degree of connection of a category or code. It is the number of links that a code maintains with other codes in the conceptual network of the corpus and captures the complexity of each code as a node that artic ulates meanings in the interpretive model. Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 7 / 22 Inter - coder reliability – calculated as a simple percentage of agreement on the presence or absence of thematic approaches and tones in each quotation – reached 100% after several rounds of discussion and refinement of the codebook. ATLAS.ti v23 was used as a host environment to manage quotations, apply codes, and visualize semantic networks. The codebook within ATLAS.ti v23 was subsequently refined by one of the coders (removing formal duplications, i.e. , gender and number markers, semantic redundancies, and merging codes) before proceeding to analyze the three parameters groundedness , co - occurrences , and density . Data V alidation Software - assisted coding based on NLP enabled the triangulation of data, the validation of the two initial categories and the emergent codes, and the visualization of semantic networks. The process was iterative and based on constant verification between codes and coded quotations. Analysis of R esults Patterns were identified through analysis of groundedness , densities , and co - occurrences to detect the strength of associations between codes, i.e., between thematic approaches and tones. To support the interpretation of the results, we established a scale for co - occurrences : <15 to indicate a weak association, 15 - 30 a moderate association, 31 - 49 a strong association, and ≥ 50 a very strong association. The scale was sized and adjusted using the category t hematic focus to avoid distortions caused by high groundedness in the category t one . Finally, to determine the significance of code co - occurrences, we established a co - occurrence threshold based on significance and noise reduction, aiming to retain only the most robust conceptual connections. Specifically, those co - occurrences detected at least 15 times and representing at least 15% in relation to the least frequent code were considered significant. ANALYSIS AND INTERPRETATION OF RESULTS The coding of the corpus revealed that journalists/media apply very different thematic approaches in their coverage of the AI Act. The coding covers 67 codes: 58 for thematic approaches and 9 for tones. Results are presented by RQ. RQ1. What categories of risks do journalists/the media emphasize? The findings span an incredibly broad range of topics, ranging from risks in general right through to philosophical issues ( Table 1 ). The category ‘ thematic approach ’ amounts to 3965 coded quotations (G), while the category ‘ journalistic tone ’ totals 3334 coded quotations (G). These totals provide the necessary denominator to interpret the specific groundedness values of the dominant codes analyzed in our study. Descriptive R esults Principal thematic approaches Risks are the dominant and most in - depth thematic approach (G = 488). Liberal newspapers ( El Español , El Mundo ) slightly prioritize them compared to progressive newspapers ( El Diario , El País ), which emphasize h uman r ights and e thics. Table 1 . Thematic approaches with higher groundedness (G) in the journalistic corpus Codes G Risks 488 Legal implications – GDPR comparison 473 Human rights – Rule of law – User protection – European values 450 Ethics – Governance 439 Social/civilizational – Changes – Impact on people ’ s daily lives 387 Evaluation – Transparency – Explainability – C ertification AI models 356 Security – Cybersecurity – Cybercrime prevention 343 Challenges – Discrepancies – Questioning – Certainty – Controversies 319 Data (collection, market, protection, privacy, and intimacy) 277 Benefits (economic, health, and climate change) – Opportunities 247 Meneses - Fernández & Chinea Rodríguez 8 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 Emphasized risks The risks represented appertain to: Information vs. disinformation ( g roundedness = 159, d ensity = 9): control of disinformation, manipulation, indistinguishability between truths/lies. Privacy (G = 277, D = 6): violation of privacy and intimacy. Biases (G = 199, D = 6) by gender, race, vulnerable persons. Economic: impact on employment (G = 135, D = 4), damage to intellectual property (G = 64). Sociopolitical (G = 93, D = 8): damage to democracy and electoral processes. Health: mental injury (G = 52). Existential - philosophical (G = 32). Significant risk co - occurrences Human, fundamental and legal rights (Co - occurrence = 59). Cybercrime/cybersecurity (C = 53). Misinformation/disinformation (C = 24). Biometry (G = 206, C = 23). Interference in democracies and electoral processes (G = 93, C = 14). Labor and employment impact (G = 135, C = 16). Interpretations Risk society and media amplification of risk The preponderance of the RBA confirms Beck ’ s (1998) thesis on the risk society and the technological threats that are subject to public debate. Newspapers portray the AI Act as an institutional response to these real risks. The predominantly critical tone, marked by expressions of concern and caution, reflects this risk society in which the media highlight the emergence of technological threats. The various approaches and tones correspond to what Kasperson et al. (1988, 2022) called “ the social amplification of risk ” , where the media articulate and amplify different dimensions of the perceived threat. The newspapers analyzed clearly highlight the risks of AI and, with that, the need to regulate it. Integrated approach The thematic approaches identified show that the Spanish newspapers take a complex and integrated approach towards AI risk, as they address not only AI ’ s technical dimensions, but also its economic, social, political, legal, cultural, and even philosophical ones as well. In line with Kasperson et al. (2022), this coverage illustrates the need for governments, regulators, experts, the media, academics, researchers, civil society, and industry to build effective public communication and the regulatory frameworks needed to manage the social amplifications of risk given their ripple effects and potential consequences. It also illustrates the importance of the media ’ s role as an arena for shaping public perceptions and fostering responsible technological development. The distribution of thematic approaches highlights the prominence given to risks - which occupies top position in the media agenda (see Table 1 ) – regardless of whether they are security risks or existential risks (McCombs & Valenzuela, 2021; Weaver et al., 2004). In counterbalance, we have the construct ‘ human rights - rule of law - protection - European values ’ , which guides protective regulations and the creation of safeguards affecting companies and individuals. The news framings confirm the predominance of risks outweighing benefits, with emphasis on attribution of responsibility (Semetko & Valkenburg, 2000). The challenges - discrepancies - questioning - certainty approach (G = 319) defines the conflict framing (Entman, 1993, 2007), reflecting tensions during the regulatory process. Hence, the need for a legal framework that mitigates conflicts of interest, liability, ethical issues, and even dystopian aspects. Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 9 / 22 RQ2 . In the newspapers, what connections emerge between public consultations & negotiations and approval of the AI Act, and the narratives surrounding the EU ’ s strategic autonomy? Descriptive R esults Presence Digital sovereignty has relatively low groundedness in the Spanish newspapers analyzed (G = 40). It stands out as the third code in connection density (D = 18), followed by: Technology corporations - lobbying (D = 34). Human rights/European values (D = 21). Significant co - occurrences of digital sovereignty Human rights/European values (C = 7). Ethics (C = 5). Risks (C = 5). Economy - competitiveness (C = 5). Geopolitical context The AI Act is represented as an historic agreement that positions the EU as a leading international regulator (G = 176), in an informative context that also mentions other AI superpowers (China G = 68, USA G = 44). Regulatory leaders pushing back against extra - European Big Tech (corporate lobbying) (G = 79, D = 34), the importance of Data (G = 277), and Security/cybersecurity (G = 343). Interpretation Paradox between theoretical relevance and media coverage Specialist literature contains numerous references to and discussions on the EU ’ s priorities, and its strategic objective of achieving digital sovereignty (Calderaro & Blumfelde, 2022; Celeste , 2021; Floridi, 2020; Pohle, 2020). However, despite its strategic importance it receives less coverage from the media. Explanatory factors for lower media coverage Conceptual ambiguity: as we indicate in the theoretical framework, digital sovereignty is an ambiguous concept and subject to different interpretations (Falkner et al., 2024), which suggests that its limited traction in the media (groundedness) stems from journalists ’ insufficient understanding of the topic. The sheer range of different meanings also makes it difficult for journalists to translate AI into plain terms for their readers. Furthermore, the AI Act does not specify who holds sovereignty, and whether it lies with the end users, companies, individual member states (Pohle & Thiel, 2020), or the EU as a whole (Celeste, 2021; Floridi, 2020). In other words, it raises the questions: who is sovereign? and who exercises sovereignty? This lack of specificity likely discourages any attempts to address the matter. Geopolitical dimension: European digital sovereignty is also reshaping geopolitical (im)balances (Bellini et al., 2025; Dammann & Glasze, 2023; Hoffmeister, 2024), which only serves to make discussions on the concept even more turbid. Journalistic trends: the tendency is for journalists to lean more towards themes with immediate impact (risks, rights, and security), which will also relegate it. Significance of co - occurrences Co - occurrences with human rights, risks, ethics, and economics reaffirm Europe ’ s goal to protect the flow of data and information being that they are valuable assets while also safeguarding individual dignity and freedom. Meneses - Fernández & Chinea Rodríguez 10 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 This aspiration is based on the stance taken by some authors (Mügge, 2024) who argue that digital sovereignty (of AI) does not limit innovation, instead it offers competitive advantages and can further other socio - economic values. RQ3 . How do journalistic narratives represent the conflict of interests between digital activism and corporate lobbying when reporting on the AI Act? The AI Act has set the stage for a discursive battleground between corporate lobbying vs. digital activism. We evidenced the representation of this conflict in the visibility, voice and treatment given to both actors during the coverage of the AI Act i . Descriptive R esults These reveal significant imbalances in coverage: Journalists pay more attention to corporate lobbying (G = 79) than to digital activism (G = 62). They appear dissociated and are not addressed in the same narratives. Corporate lobbying shows the highest density in the study (D = 34). Digital activism has a considerably lower density (D = 2). Significant co - occurrences of corporate lobbying Challenges - contradictions (C = 9). AI risks (C = 9). Data - market data (C = 8). Evaluation of AI systems (C = 7). Significant co - occurrences of digital activism Human r ights (C = 8). Interference in democracy/electoral processes (C = 6). Ethical/governance concerns (C = 6). Significance of major co - occurrences The high co - occurrences of corporate lobbying and digital activism with certain aspects of the AI Act in the coding reveal remarkable interpretive patterns. Lobbying and obstacles - challenges - contradictions (C = 9) indicate that journalistic narratives link technology corporations with obstacles and contradictions arising directly from the AI Act. Newspapers portray Big Tech as key players when it comes to pushing back against regulatory proposals, which raises questions about the viability or effectiveness of such regulation. Lobbying and AI risks (C = 9) indicate that the very power held by Big Tech highlights the importance of regulating AI. Lobbying and data - market (C = 8), which relate to the previous dataset, reflect the narrative link between corporate interests and the control/exploitation of data; that is, corporations lobby to protect their economic interests in the data market; hence, they lobby to preserve their business model which is based on access to and the commercial use of user data. Lobbying and evaluation of AI systems (C = 7) are indicative of both corporate interest in influencing the standards they will have to meet in the EU, and the urgency of legally addressing the potential risks posed by AI. Digital activism and human rights (C = 8) show that newspapers link both approaches. Activist organizations are portrayed as guardians of rights and as an ethical - legal counterweight to the abusive development and use of AI. Activism and interference in democracy/electoral processes (C = 6) is another significant co - occurrence because it alludes to concerns about the impact of AI on democratic processes. The newspapers represent Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 11 / 22 activist organizations as voices that warn of the risk of information being manipulated, the spread of disinformation, and distortion in electoral processes. Activism and ethical concerns/governance (C = 6) call for ethical frameworks and governance structures for AI. In their role as technology watchdogs, activist NGOs promote ethical principles that guide a restrictive AI Act, demanding transparency and the use of rigorous standards when evaluating AI technologies. M edia R epresentation of the A ctors Digital rights organizations by newspaper Table 2 summarizes the differential treatment given to digital rights organizations in each newspaper. Representation of Big Tech The comparative analysis of mentions of technology companies shows differences by newspaper ( Table 3 ). Interpretation Relational interpretation of coverage The comparison between the two actors touches on two areas: one is their influence on the drafting of the AI Act; the other, and the one that interests us here, the Spanish media ’ s coverage of each actor. The analysis reveals significant patterns in the portrayal of Big Tech, as well as tensions between innovation and regulation: - Polarization, as journalistic attention is focused on a handful of corporations highlighted within the media discourse (OpenAI, Google, Microsoft). - Tension, as the narrative oscillates between technological innovation and regulatory implications, with companies taking opposing stances – those in favor (IBM) and those against (Meta, OpenAI) to regulation. - Localization of the tension in the EU, a territory where conflicts arise between the geopolitical interests of Big Tech and the AI Act. - Socioeconomic transformation, as newspapers highlight labor and business tensions linked to corporate governance of AI; in fact, El Diario reveals that only one - third of companies apply ethical practices in their use of AI. Therefore, we observe regulatory, geopolitical, labor and strategic - corporate tensions that affect users ’ rights; in short, tension between technology - power. Ciuriak (2018) summarizes this situation in terms of corporate control over vast amounts of digital data (G = 277). The reason this would be advantageous is because it grants Big Tech power that transcends the economic sphere; freedom to act without restraints would allow them to consolidate their position and influence in the market. These actions would consist of Table 2 . Digital rights organizations with/without voice by each Spanish newspaper Newspaper Digital rights organizations mentioned Main focus Featured topics Coverage intensity El Diario EDRi, Access Now, AlgorithmWatch, Citizen AI Coalition Critical of AI regulation Facial recognition, surveillance, u se of AI in social support, d iscrimination High (multiple mentions with direct quotes) El Español No direct mention Institutional approach in EU legislation Digital Services Act (DSA), Digital Markets Act (DMA), h uman - centric approach to digitalization Low (no specific mention of NGOs) El Mundo Association for Progressive Communications (APC), Human Rights Watch, Amnesty International, Digital Rights Defence Network (R3D) Human rights impact analysis Surveillance, f acial recognition, AI discrimination, Toronto Declaration Medium (contextual mentions without many direct quotations) El País EDRi (focus on Ella Jakubowska), Global Witness, Amnesty International Progress and setbacks in regulation Facial recognition, b iometric data tracking, AI threats to human rights Average (specific mentions with direct quotes) Meneses - Fernández & Chinea Rodríguez 12 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 offering access to content, capturing users whose preferences can be tracked, but they would also go hand in hand with impingements upon their rights; hence their interest in intervening in AI regulatory policies (Ciuriak, 2018; Hine, 2024). Interpretative dimensions of digital activism - Discursive positioning, as newspapers portray NGOs as watchdogs and overseers of digital rights during the negotiation of the AI Act. - Narrative polarization, given that newspapers present the hardline stance of NGOs in juxtapose with a more laisser - fair stance from other stakeholders ( “ very diluted approach ” , “ absolute minimum for companies ” El Diario , November 28, 2023). - An evolution of positions, as the newspapers show the veerings and contradictions among NGOs during the regulatory process, ranging from initially proactive activism ( “ they ask for increased protection ” El Diario , May 17, 2023) to eventual disappointment ( “ surprised and disappointed ” El País , December 9 - 10, 2023). The contrast between “ totally delighted with the text ” and “ surprised and Table 3 . Comparison of mentions of Big Tech by each Spanish newspaper Company El Diario El Español El Mundo El País OpenAI Impact of ChatGPT on the market, c ontract with Microsoft, EU p rivacy i nvestigations, p ossible withdrawal of ChatGPT from Europe Significant impact of ChatGPT, n on - compliance of GPT - 4 with EU requirements, r egulatory and transparency debates Key player in language models, p ossible abandonment of Europe by EU regulations, c losed - source model Collaboration with Microsoft, i ntegration of technologies in Microsoft products Google Bard development (not available in Europe), AI tools for advertising, c ompetition with ChatGPT Sundar Pichai ’ s position on regulation, e mphasis on potential benefits (health), w arnings about risks (misinformation) Development of Gemini, c ompetition with OpenAI and Microsoft, i nvestment in AI startups Launch of Gemini (multimodal AI), m erger of DeepMind and Google Brain, m arket positioning after impact of OpenAI Microsoft “ Copilots ” to automate tasks, t ransforming office work, OpenAI technology integration in Bing, Satya Nadella advocates for global standards Development of standards for r esponsible AI, Natascha Crampton as c hief r esponsible AI o fficer, c ode of c onduct with ethical values Major investor in AI, p artnership with OpenAI, Investment strategies towards AI (no metaverse) Brad Smith advocates responsible regulation, s ignificant investment in Spain, Integration of AI in products (Bing, CoPilot) Goal Advances in AI for p rotein s tructures, a pplications in m edical d iagnostics Llama model does not meet EU requirements, r egulatory c hallenges Conflicts with European regulation, n o launch of new models in Europe, Llama2 open - source development No offer of generative AI models in Europe, w aiver of using user data for training IBM 8 , 000 layoffs due to automation, HR downsizing, 40% of large companies in Spain use AI. Only 36% with ethical policies Concern about exclusion in European projects, s upport for risk - based approach to AI Act, o pposition to overly detailed regulation Not mentioned Christina Montgomery emphasizes smart regulation, Watsonx.governance development, f ocus on ethical development Apple Competition in virtual assistants, r elative failure of Alexa (Amazon) Not mentioned Delayed launch of AI in Europe, r estricted use of ChatGPT among employees Not mentioned Amazon Competition in virtual assistants, Alexa failed to live up to expectations Development of codes of conduct for AI Investment in AI development, i nvestment in related startups Not mentioned Nvidia Not mentioned Not mentioned Key supplier of AI hardware Not mentioned Intel Not mentioned Not mentioned Not mentioned Support for AI regulation, r esponsible platform development, Greg Lavender pro AI accessible to everyone Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 13 / 22 disappointed ” from the same organization (EDRi) reflects the complexity of activist positions on the various drafts and elements of the AI Act. - Legitimacy and expertise, notwithstanding the above, journalists underscore the track record, specialization ( “ coordinator of the facial biometrics program ” ) and institutional role ( “ head of public policy ” ) of NGOs, thereby reinforcing their legitimacy to monitor and critically evaluate the AI Act. RQ4. Which journalistic tones are associated with the Spanish media ’ s coverage of the AI Act? Our analysis reveals predominant discursive patterns and non - observance of certain aspects surrounding the AI Act that deserve attention. Descriptive R esults We identified nine tones in the coverage by the Spanish media ( Table 4 ). Critical and positive tones predominate, followed by reflexive - propositional - speculative. This distribution shows that AI Act coverage is complex and multidimensional ( Table 5 ). Significant C o - O ccurrences We analyze the most frequent associations with the greatest comparative and interpretative interest. The critical tone The critical tone appearing with the highest frequency is mainly associated with: Ethical concerns (C = 138) Table 4 . Tones of journalistic treatment and groundedness (G) Tone: Codes G Critical 1 , 094 Positive 1 , 056 Reflective – Propositional – Speculative on AI uses 911 Concern – Warning – Alarming 690 Cautious – Responsible 450 Informative – Neutral 241 Negative 212 Analytical – Skeptical 155 Educational – Explanatory 49 Table 5 . Significant co - occurrences (C) between tones and thematic approaches Tone Thematic f ocus C Critical Ethical concerns 138 Challenging the AI Act 133 Human rights 132 Risks 125 Algorithmic bias – D iscrimination 110 Positive Miscellaneous benefits 158 Human rights 128 Civilizational – S ocial changes 111 Historic agreement 95 Security/cybersecurity 92 Public/private, international cooperation 89 Ethics – G overnance 83 Concern – W arning – A larming Risks 135 Human rights 84 Civilizational – S ocial changes 72 Security/cybersecurity 70 Cautious – R esponsible Ethics – G overnance 60 Risks 93 Security – C ybersecurity 60 Civilizational – S ocial changes 55 Ethics – G overnance 54 Meneses - Fernández & Chinea Rodríguez 14 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 Challenging the AI Act (C = 133) Human rights (C = 132) Risks (C = 125) Algorithmic bias - discrimination (C = 110) Positive tone is associated with Miscellaneous benefits (C = 158) Human rights (C = 128) Civilizational - social changes (C = 111) Historic agreement (C = 95) Security/cybersecurity (C = 92) Public/private, international cooperation (C = 89) Ethics - governance (C = 83) The concerned - warning - alarming tone The concerned - warning - alarming tone is associated with: Risks (C = 135) Human rights (C = 84) Civilizational - social changes (C = 72) Security/cybersecurity (C = 70) Ethics - governance (C = 60) The cautious - responsible tone The cautious - responsible tone is associated with: Risks (C = 93) Security/cybersecurity (C = 60) Civilizational - social changes (C = 55) Ethics - governance (C = 54) Significance of C o - O ccurrences In the associations between tones and themes, we observe discursive patterns that require interpretation. Criticism and surveillance of power The frequency of association between critical tone and ethical concerns, human rights and risks is by no means incidental. This pattern reflects that newspapers are playing their role in monitoring, confronting, and holding those in power to account ( Carlson, 2018). And, that they do so with an attitude that is critical towards possible shortcomings or threats when dealing with these sensitive issues. Regulatory uncertainty The significant correlation between the concern - warning - alarming tone and the contradictions - uncertainties thematic focus (C = 45) is evidence of skepticism & questions regarding the effectiveness of the AI Act. Paul (2024) notes that these doubts are intensified by the rapidly changing technological landscape and pressures from global economic competition increasing competitiveness. Interpretation of R esults Critical - positive duality The predominance of a critical tone is key when linked to questioning the effectiveness of the prohibitions laid out by the AI Act ( a rt. 5), and the effectiveness of supranational governance mechanisms ( c hapter III). At Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 15 / 22 the same time, the widespread use of a positive tone reflects recognition of regulatory progress within the EU. This dual narrative is consistent with similar findings by other researchers (Brennen et al., 2021). In fact, this discursive duality is characterized in the literature as a recurrent phenomenon in communication about incipient technological policies, with simultaneous narratives on both progress and caution (e.g. , Watson et al., 2024). For Jobin et al. (2019, p. 396), it reflects societal ambivalence toward AI: “ Similar contrasts emerge between avoiding harm at all costs and the perspective of accepting some degree of harm as long as risks and benefits are weighed against each other. ” Thus, newspapers simultaneously celebrate innovation whilst also scrutinizing their regulatory frameworks, reflecting the underlying tensions that are present in the face of technological progress. Therefore, we agree with Tartaro (2024, p. 11), who states: “ Future regulatory frameworks must strive for a balanced approach that fosters innovation while mitigating risks and ensuring that AI development and deployment adheres to societal values [...] and European values ” . Reflective - propositional - speculative tones on AI uses, and tones of concern and caution The associations between these tones and various thematic approaches reflects concern about the actual scope of the AI Act, particularly when they question its effectiveness. The notable presence of the reflexive - propositional - speculative tone relates to future developments in AI. Here, the newspapers point to the need to regulate new risks. This reaction, already noted by Zuboff (2022), is caused by a regulatory framework that does not seem to satisfy stakeholders ’ expectations (NGOs and Big Tech). Educational - explanatory deficit It is striking how few associations there are between the infrequently used educational - explanatory tone (G = 49) and the three thematic approaches from our RQs. This neglect of journalism ’ s formative role sits in contrast with the provisions of the AI Act that call for greater digital literacy. The coverage of complex regulations and technology, especially when cause for concern, is also weak. This finding contrasts with the provisions of the AI Act ( r ecitals 20, 56, 91, 165, a rticles. 3.56, 4, 66.e.f, and 95.2.c) (EU, 2024), which stresses the importance of promoting the critical and introspective use AI. If we look at the thematic approaches that have emerged as a whole, the plausible interpretation is that newspapers bring more sensational elements to the forefront, whilst relegating digital literacy to the backseat, even though Frau - Meigs (2024, p. 525) points out “ It showed that m edia and i nformation l iteracy can support algorithm literacy for the general users, including teachers and journalists focused on the fight against disinformation and in favor of quality information and search ” . We summarize our main findings from Spanish media coverage of the AI Act in Figure 2 . Figure 2 . Conceptual map of extended semantic fields in AI regulation summarizing findings on Spanish media coverage of the AI Act (the a uthors ’ own elaboration) Meneses - Fernández & Chinea Rodríguez 16 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 DISCUSSION Our results and findings have enabled us to demonstrate the complexity of media coverage of the AI Act by Spanish newspapers. Integration of T hemes and T ones The results cover both thematic approaches and journalistic tones. The risks of AI lay in the salient features, which justify the promulgation of the AI Act. The innumerable benefits offered by this technology chart the ambivalence that is characteristically found in a society weighing up the costs & benefits of technological advancements (Jobin et al., 2019). Identifying thematic approaches and journalistic tones surrounding the AI Act provided insights into the process of constructing meaning through journalistic content, which, according to Scheufele and Lewenstein (2005), citizens will use to form their heuristics - in this instance, regarding the regulation of AI. Narrative C omplexity and P otential C onfusion The findings reveal a narrative panorama characterized by a complex thematic agenda featuring a variety of realistic tones. This complexity, however, can generate confusion and lead to skepticism among readers, as detected by Brossard (2013). The co - occurrences show that the dissemination of the AI Act appears to be shaped by a discourse dominated by warnings of risks and threats. Consistently, we identify frames that emphasize the restrictive and protective nature of the AI Act. However, one crucial element is missing: the path to achieving EU digital sovereignty. This is an aspiration in line with the claims of pro - rights digital activism, despite there being the greater narrative presence of corporations (Algorithm Watch, 2023; EDRi, 2023). During the regulatory process, digital activism is presented as a more horizontal phenomenon (Bennett & Segerberg, 2012), while corporate lobbying is associated with vertical and pragmatic actions. As Sharon and Gellert (2024) warn, these tactics seek to influence regulation and even circumvent it if they are not approached with foresight. Activism and lobbying are not presented to readers side by side in the same text, coverage leans quantitatively towards the latter. The P edagogical D eficit in N ews C overage The journalists do not combine reporting on the AI Act with the role of educating and raising public awareness. This lack of comprehensive narratives and pedagogical weakness exposes the public to technical critiques and news reports that lack sufficient guidance on how to understand highly complex pieces of legislation and technology. Risk theory (Beck, 1998; Kasperson, 1992; Kasperson et al., 1988, 2022) states that this dissociation reflects the lack of a holistic approach that connects critical thinking, media literacy, and awareness; and it shows the selective amplification of aspects that are, perhaps, examples of more titillating journalism (risk/human rights, criticism/opposition). Furthermore, we have identified lexical patterns that reveal how the AI Act and its ethical, legal, economic, political, and social implications are portrayed (Brause et al. , 2023; Eubanks , 2018; Noble , 2018). The pattern consists of at times messy and convoluted techno - political vocabulary offering fragmented accounts of risk, digital sovereignty, and activism - corporate tensions. Spanish journalists have employed language that blends technical terms with political, legal, and ethical concepts, reflecting the tensions between regulations and innovation. However, there is a lack of effort to translate this thematic complexity to assist the reader. In other words, our findings show that the Spanish media have assumed a pivotal role in informing the public about a regulatory paradigm without shying away from its complexities, but it lacks adequate efforts to educate readers. CONCLUSIONS Risks ( RQ1 ), the central protagonist. Spanish newspapers place the spotlight firmly on technological risks in their coverage of the AI Act, portraying it as the first major attempt to regulate a technology with incredible Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 17 / 22 benefits but also extreme risks. Journalists cover the full range of risks, from the biometric tracking of individuals through to the undermining of European humanist values. This is the main narrative thread. European digital sovereignty ( RQ2 ): insufficient coverage. Despite its strategic relevance for the EU, it lacks notable depth in journalistic discourse. This finding is surprising from a European perspective; however, this may be due to it being a complex and ambiguous concept intertwined with human rights, ethics and risks. Although the AI Act is portrayed as a significant international landmark, it does not help to highlight the strategic importance of achieving digital sovereignty. The Spanish newspapers have tended to focus on more sensational topics (risks, security, and rights), limiting exploration of who exercises this sovereignty and how it affects different stakeholders. As yet, there is no consensus on the concept itself, which generates ambiguity and constrains its scope in journalistic discourse, despite its impacts being widespread. Tensions represented: regulation - innovation vs. activism - lobbying ( RQ3 ). We identified the regulation - innovation tension in the juxtaposition between pro - rights NGOs vs. corporate lobbying, a recurring theme in journalistic narratives accentuated by news about challenges, discrepancies and uncertainty in the negotiations over the AI Act. The newspapers have extensively covered the dilemma of regulation over innovation (or rights - development), which has been resolved through legislation aimed at balancing social protections and technological development in the EU. They portray NGOs as legitimate watchdog stakeholders; however, our analysis reveals an imbalance in coverage: corporations with economic interests in AI receive more attention than organizations that are pro - digital rights. This suggests that the media coverage is skewed towards the relationship Big Tech - EU regulation , favoring lobbying perspectives over voices advocating for digital rights. As such, this would be an opportune moment to call for less bias and more balanced coverage, as well as greater input from civil society. Variety of journalistic tones ( RQ4 ). This indicates that the AI Act went through public consultations and negotiations in which different stakeholders competed to define the dominant narrative surrounding a piece of legislation that transcends EU borders. The predominance of the critical tone, followed by positive and reflective elements, indicates that newspapers exercise the role of watchdog inside the complex landscape of EU laws, regulations, and decision - making processes. They thus contribute to a nuanced public debate, rather than to an oversimplistic ‘ pro/anti ’ or ‘ positive/negative ’ polarization. This interpretation is consistent with previous research on media coverage of emerging technologies (Brossard, 2013; Scheufele & Lewenstein, 2005), which identifies similar patterns of reflexive ambivalence in the coverage of new technological regulations. Taken together, the newspapers represent the AI Act as a historic landmark for non - European regulatory initiatives. Throughout their coverage, they have developed a conceptual framework for addressing this regulation that integrates technical, ethical, economic and political dimensions. The lexical usages cement the narratives, representations and framings of the legislation (AI Act) and of the technology being regulated (AI) around three discursive vectors: technological risks, protection of rights, and technological innovation. However, the complex informative panorama observed is not balanced out by media literacy or an educational - explanatory tone. Finally, we believe that the contribution made by this study lies in its practical application. It assists our understanding of how the legitimacy of AI policies is constructed, reveals narrative patterns and biases in the media ’ s portrayal of complex regulatory processes, and provides insights into a crucial issue that, as yet, has scarcely been explored in the literature, especially from a European perspective, this being: AI regulation as reported on and portrayed in the media. Limitations of the S tudy The AI Act is just the first step in the journey towards regulation and towards a growing body of research examining the media narratives that will accompany it. We openly acknowledge the limitations of our study, which focuses on the sender of the media message. Naturally, the research horizon must expand in several complementary directions. First, it must expand towards an understanding of how different audiences receive and interpret this coverage. Second, it should explore the role of the media in regulatory diplomacy, both for the AI Act and for new legislation being drafted in other regions. These new studies would allow us to identify Meneses - Fernández & Chinea Rodríguez 18 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 patterns in how different media landscapes, from diverse socio - cultural contexts, narrate the regulation of this technology. At a time when AI is reshaping the balance of power, it is essential to look beyond the timeframe of the AI Act. Only then will we be able to understand whether or not the media are engaging in regulatory media diplomacy; and whether different points in time and different geographical contexts, with their own distinct regulatory processes, are shaping disparate media narratives. Author contributions: MDM - F: conceptualization, data curation, formal analysis, investigation, methodology, validation, writing – original draft , w riting – review & editing ; MSCR: data curation, investigation, validation, w riting – review & editing . Both authors approve d the final version of the article. Funding: The authors received no financial support for the research and/or authorship of this article. Acknowledgments: The authors would like to thank the reviewers and editors for their constructive observations, which have significantly improved the article . Ethics declaration: This study was based on news articles published in leading Spanish national newspapers, as well as on literature obtained from public bibliographic databases, and reports available on official European Union websites. It did not involve human participants, animals, or personal data; therefore, institutional ethical approval was not required . AI statement: The authors declare that no generative AI tools were used in the preparation of this manuscript, including for text writing, editing, or figure production . Declaration of interest: The a uthors declare d no competing interest. Data availability: Data generated or analy z ed during this study are available from the authors on request. REFERENCES AI - HLEG. (2019a). A definition of artificial intelligence : Main capabilities and scientific disciplines. European Commission . https://digital - strategy.ec.europa.eu/en/library/definition - artificial - intelligence - main - capabilities - and - scientific - disciplines AI - HLEG. (2019b). Policy and investment recommendations for trustworthy artificial intelligence . European Commission . https://digital - strategy.ec.europa.eu/en/library/policy - and - investment - recommendations - trustworthy - artificial - intelligence Algorithm Watch. (2023). EU legislators must close dangerous loophole in AI Act, Statement, 7.09.2023. Algorithm Watch . https://algorithmwatch.org/en/eu - must - close - loophole - ai - act/ Allaham, M., Kieslich, K., & Diakopoulos, N. (2025). Global perspectives of AI risks and harms: Analyzing the negative impacts of AI technologies as prioritized by news media . arXiv. https://doi.org/10.48550/arXiv.2501. 14040 Beck, U. (1998). La sociedad del riesgo. Hacia una nueva modernidad [The risk society. Towards a new modernity]. Paidós. Bellini, P., Campanella, E., & Piccioni, A. (2025). The impact of artificial intelligence and data on political power and geopolitical equilibria. In G. R. Marseglia, P. Previtali, & A. Reali (Eds.), Socio - economic impact of artificial intelligence. A European management perspective (pp. 15 - 22). Springer. https://doi.org/10.1007/ 978 - 3 - 031 - 73514 - 1_3 Bennett, W. L., & Segerberg, A. (2012). The logic of connective action: The personalization of contentious politics. Information, Communication & Society, 15 (5), 739 - 768. https://doi.org/10.1080/1369118X.2012. 670661 Brause, S. R., Zeng, J., Schäfer, M., & Katzenbach, C. (2023). Media representations of artificial intelligence: Surveying the field. In S. Lindgren (Ed.), Handbook of critical studies of artificial intelligence (pp. 277 - 288). Edward Elgar Publishing. https://doi.org/10.4337/9781803928562.00030 Brennen, J. S., Howard, P. N., & Nielsen, R. K. (2021). What to expect when you ’ re expecting robots: Futures, expectations, and pseudo - artificial intelligence in UK news. Journalism, 22 (5), 1287 - 1305. https://doi.org/ 10.1177/1464884920947535 Brossard, D. (2013). New media landscapes and the science information consumer. PNAS, 110 (supplement_3), 14096 - 14101. https://doi.org/10.1073/pnas.1212744110 Buozis, M., & Creech, B. (2017). Reading news as narrative. A genre approach to journalism studies. Journalism Studies, 19 (10), 1430 - 1446. https://doi.org/10.1080/1461670X.2017.1279030 Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 19 / 22 Burscher, B., Vliegenthart, R., & de Vreese, C. H. (2016). Frames beyond words: Applying cluster and sentiment analysis to news coverage of the nuclear power issue. Social Science Computer Review, 34 (5), 530 - 545. https://doi.org/10.1177/0894439315596385 Calderaro, A. , & Blumfelde, S. (2022) Artificial intelligence and EU security: T he false promise of digital sovereignty . European Security, 31 ( 3 ) , 415 - 434 . https://doi.org/10.1080/09662839.2022.2101885 Carlson, M. (2018). The i nformation p olitics of j ournalism in a p ost - t ruth a ge. Journalism Studies, 19 (13), 1879 - 1888. https://doi.org/10.1080/1461670X.2018.1494513 Celeste, E. (2021). Digital sovereignty in the EU: Challenges and future perspectives. In F. Fabbrini, E. Celeste, & J. Quinn (Eds.), Data protection beyond borders: Transatlantic perspectives on extraterritoriality and sovereignty (pp. 211 - 228). Bloomsbury Publishing. https://doi.org/10.5040/9781509940691.ch - 013 Ciuriak, D. (2018). The economics of data: Implications for the data - driven economy. In CIGI (Ed.), Data governance in the digital age . Centre for International Governance Innovation. https://doi.org/10.2139/ ssrn.3118022 Couture, S., & Toupin, S. (2019). What does the notion of “sovereignty” mean when referring to the digital? New Media & Society, 21 (10), 2305 - 2322. https://doi.org/10.1177/1461444819865984 D ’ Angelo, P. (2019). Framing theory and journalism. In T. P. Vos, & F. Hanusch (Eds.), The international encyclopedia of journalism studies (pp. 1 - 10). Wiley. https://doi.org/10.1002/9781118841570.iejs0021 Dammann, F., & Glasze, G. (2023) Governing digital circulation: The quest for data control and sovereignty in Germany. Territory, Politics, Governance, 11 (6), 1100 - 1120. https://doi.org/10.1080/21622671.2022. 2141850 d e Almeida, P. G. R., d os Santos, C. D., & Farias, J. S. (2021). Artificial intelligence regulation: A framework for governance. Ethics and Information Technology, 23 (3), 505 - 525. https://doi.org/10.1007/s10676 - 021 - 09593 - z EC. (2020). Artificial intelligence – Ethical and legal requirements. European Commission . https://ec.europa.eu/ info/law/better - regulation/have - your - say/initiatives/12527 - Artificial - intelligence - ethical - and - legal - requirements_en EC. (2021). Artificial intelligence – Ethical and legal requirement. Public consultation. European Commission . https://ec.europa.eu/info/law/better - regulation/have - your - say/initiatives/12527 - Artificial - intelligence - ethical - and - legal - requirements_en EC. (2024). European approach to artificial intelligence. European Commission . https://digital - strategy.ec.europa.eu/en/policies/european - approach - artificial - intelligence EDRi. (2023). EU trilogues: The AI Act must protect people ’ s rights, statement, 12.07.2023. EDRi . https://edri.org/wp - content/uploads/2023/07/Civil - society - AI - Act - trilogues - statement.pdf Entman, R. M. (1993). Framing: Toward clarification of a fractured paradigm. Journal of Communication, 43 (4), 51 - 58. https://doi.org/10.1111/j.1460 - 2466.1993.tb01304.x Entman, R. M. (2007). Framing bias: Media in the distribution of power. Journal of Communication, 57 (1), 163 - 173. https://doi.org/10.1111/j.1460 - 2466.2006.00336.x EU. (2024). Regulation (EU) 2024/1689 of the European Parliament and of the Council of 24 April 2024 laying down harmonised rules on artificial intelligence (Artificial Intelligence Act) and amending certain Union legislative acts. Document 32024R1689. Official Journal of the European Union, L 2024/1689 . http://data.europa.eu/eli/reg/2024/1689/oj Eubanks, V. (2018). Automating inequality: How high - tech tools profile, police, and punish the poor . St. Martin ’ s Press. Falkner, G., Heidebrecht, S., Obendiek, A., & Seidl, T. (2024). Digital sovereignty – Rhetoric and reality. Journal of European Public Policy, 31 (8), 2099 - 2120. https://doi.org/10.1080/13501763.2024.2358984 Feldstein, S. (2024). Evaluating Europe ’ s push to enact AI regulations: How will this influence global norms? Democratization, 31 (5), 1049 - 1066. https://doi.org/10.1080/13510347.2023.2196068 Fleming, S. (2025). What is digital sovereignty and how are countries approaching it? World Economic Forum . https://www.weforum.org/stories/2025/01/europe - digital - sovereignty/ Floridi, L. (2020). The fight for digital sovereignty: What it is, and why it matters, especially for the EU. Philosophy & Technology, 33 , 369 - 378. https://doi.org/10.1007/s13347 - 020 - 00423 - 6 Frau - Meigs, D. (2024). Algorithm literacy as a subset of media and information literacy: Competences and design considerations. Digital, 4 , 512 - 528. https://doi.org/ 10.3390/digital4020026 Meneses - Fernández & Chinea Rodríguez 20 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 Gasser, U. (2023). An EU landmark for AI governance. Science, 380 (6651), Article 1203. https://doi.org/10.1126/ science.adj1627 Gourlet, P., Ricci, D., & Crépel, M. (2024). Reclaiming artificial intelligence accounts: A plea for a participatory turn in artificial intelligence inquiries. Big Data & Society, 11 (2). https://doi.org/10.1177/ 20539517241248093 Grant, P. (1983). Technological sovereignty: Forgotten factor in the ‘ hi - tech ’ razzamatazz. Prometheus, 1 (2), 239 - 270. https://doi.org/10.1080/08109028308628930 Hayles, N. K. (1999). How we became posthuman: Virtual bodies in cybernetics, literature, and informatics . The University of Chicago Press. https://doi.org/10.7208/chicago/9780226321394.001.0001 Helberger, N. (2024). FutureNewsCorp, or how the AI Act changed the future of news. Computer Law & Security Review, 52 , Article 105915. https://doi.org/10.1016/j.clsr.2023.105915 Helberger, N., & Diakopoulos, N. (2023). The European AI Act and how it matters for research into AI in media and journalism. Digital Journalism, 11 (9), 1751 - 1760. https://doi.org/10.1080/21670811.2022.2082505 Hine, E. (2024). The Brussels effect is no match for the might of big tech: US AI regulation is being shaped by lobbyists in state after state. Nature, 633 , Article S15. Hoffmeister, K. (2024). The dawn of regulated AI: Analyzing the European AI act and its global impact. Zeitschrift für Europarechtliche Studien, 27 (2), 182 - 212. https://doi.org/10.5771/1435 - 439X - 2024 - 2 - 182 Hopmann, D. N., Vliegenthart, R., De Vreese, C., & Albæk, E. (2010). Effects of election news coverage: How visibility and tone influence party choice. Political Communication, 27 (4), 389 - 405. https://doi.org/10.1080 /10584609.2010.516798 Horowitz, M. C., Allen, G. C., Saravalle, E., Cho, A., Frederick, K., & Scharre, P. (2018). Artificial intelligence and international security . Center for a New American Security. Jacobs, L. (2020). Tone and threats on television over time: A longitudinal analysis of news about Roma in Flanders (2003 - 2017). International Journal of Communication, 14 (2020), 46 - 68. Jobin, A., Ienca, M., & Vayena, E. (2019). The global landscape of AI ethics guidelines. Nature Machine Intelligence, 1 , 389 - 399. https://doi.org/10.1038/s42256 - 019 - 0088 - 2 Kasperson, R. E. (1992). The social amplification of risk: Progress in developing an integrative framework. In S. Krimsky, & D. Golding (Eds.), Social theories of risk (pp. 153 - 178). Praeger. Kasperson, R. E., Renn, O., Slovic, P., Brown, H. S., Emel, J., Goble, R., Kasperson, J. X., & Ratick, S. (1988). The social amplification of risk: A conceptual framework. Risk Analysis, 8 (2), 177 - 187. https://doi.org/10.1111/ j.1539 - 6924.1988.tb01168.x Kasperson, R. E., Webler, T., Ram, B., & Sutton, J. (2022). The social amplification of risk framework: New perspectives. Risk Analysis, 42 (7), 1367 - 1380. https://doi.org/10.1111/risa.13926 Koziner, N. S. (2013). Antecedents and bases of framing theory in communication. Austral Comunicación, 2 (1), 1 - 25. Lane, L. (2022). Clarifying human rights standards through artificial intelligence initiatives. International and Comparative Law Quarterly, 71 (4), 915 - 944. https://doi.org/10.1017/S0020589322000380 McCombs, M., & Valenzuela, S. (2021). Setting the agenda: Mass media and public opinion (3rd ed.). Polity Press. Mügge, D. (2024). EU AI sovereignty: For whom, to what end, and to whose benefit? Journal of European Public Policy, 31 (8), 2200 - 2225. https://doi.org/10.1080/13501763.2024.2318475 Nguyen, D., & Hekman, E. (2022). The news framing of artificial intelligence: A critical exploration of how media discourses make sense of automation. AI & SOCIETY, 39 , 437 - 451. https://doi.org/10.1007/s00146 - 022 - 01511 - 1 Noble, S. U. (2018). Algorithms of oppression: How search engines reinforce racism . New York University Press. Ørmen, J., & Gregersen, A. (2019). News as narratives. In M. Powers (Ed.), Oxford research encyclopedia of communication . Oxford Academic. https://doi.org/10.1093/acrefore/9780190228613.013.908 Ouchchy, L., Coin, A., & Dubljević, V. (2020). AI in the headlines: The portrayal of the ethical issues of artificial intelligence in the media. AI & Society, 35 (4), 927 - 936. https://doi.org/10.1007/s00146 - 020 - 00965 - 5 Pagallo, U. (2023). Dismantling four myths in AI & EU law through legal information ‘ about ’ reality. In H. Sousa, P.M. Freitas, A.L. Oliveira. C. Martins, E. Vaz, & L. Barreto (Eds.), Multidisciplinary perspectives on artificial intelligence and the law (pp. 251 - 261). Springer. https://doi.org/10.1007/978 - 3 - 031 - 41264 - 6_13 Online Journal of Communication and Media Technologies, 2026 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 21 / 22 Paul, R. (2024). European artificial intelligence “ trusted throughout the world ” : Risk - based regulation and the fashioning of a competitive common AI market. Regulation & Governance, 18 (4), 1065 - 1082. https://doi.org/10.1111/rego.12563 Pehlivan. C. N. (2024). The EU Artificial Intelligence (AI) Act: An introduction. Global Privacy Law Review, 5 (1), 31 - 42. https://doi.org/10.54648/gplr2024004 Pohle, J., & Thiel, T. (2020). Digital sovereignty. Internet Policy Review, 9 (4). https://doi.org/10.14763/2020.4.1532 Pollach, I., & Hansen, L. V. (2021). Tone variation in financial news: A comparison of companies, journalists, and financial analysts. European Journal of Communication, 36 (5), 511 - 526. https://doi.org/10.1177/ 0267323121999524 Porlezza, C. (2023). Promoting responsible AI: A European perspective on the governance of artificial intelligence in media and journalism. Communications, 48 (3), 370 - 394. https://doi.org/10.1515/commun - 2022 - 0091 Porlezza, C. (2024). The datafication of digital journalism: A history of everlasting challenges between ethical issues and regulation. Journalism, 25 (5), 1167 - 1185. https://doi.org/10.1177/14648849231190232 Saperas, E. (1987). Los efectos cognitivos de la comunicación de masas: Las recientes investigaciones en torno a los efectos de la comunicación de masas: 1970 - 1986 [ The cognitive effects of mass communication: Recent research on the effects of mass communication: 1970 - 1986 ] . Ariel. Scherrer, A. (2023). How media coverage of technologies affects public opinion: Evidence from alternative fuel vehicles in Germany. Environmental Innovation and Societal Transitions, 47 , Article 100727. https://doi.org/ 10.1016/j.eist.2023.100727 Scheufele, D. A., & Lewenstein, B. V. (2005). The public and nanotechnology: How citizens make sense of emerging technologies. Journal of Nanoparticle Research, 7 , 659 - 667. https://doi.org/10.1007/s11051 - 005 - 7526 - 2 Semetko, H. A., & Valkenburg, P. M. (2000). Framing European politics: A content analysis of press and television news. Journal of Communication, 50 (2), 93 - 109. https://doi.org/10.1111/j.1460 - 2466.2000. tb02843.x Sharon, T., & Gellert, R. (2024). Regulating Big Tech expansionism? Sphere transgressions and the limits of Europe ’ s digital regulatory strategy. Information, Communication & Society, 27 (15), 2651 - 2668. https://doi.org/10.1080/1369118X.2023.2246526 Sparks, J. F., & Hmielowski, J. D. (2022). At the extremes: Assessing readability, grade level, sentiment, and tone in US media outlets. Journalism Studies, 24 (1), 24 - 44. https://doi.org/10.1080/1461670X.2022.2142646 Strauss, A., & Corbin, J. (2016). Bases de la investigación cualitativa: Técnicas y procedimientos para desarrollar la teoría fundamentada [Foundations of qualitative research: Techniques and procedures for developing grounded theory]. Antioquia. Tartaro, A. (2024). Value - laden challenges for technical standards supporting regulation in the field of AI. Ethics and Information Technology, 26 , Article 72. https://doi.org/10.1007/s10676 - 024 - 09809 - y Tolmeijer, S., Kneer, M., Sarasua, C., Christen, M., & Bernstein, A. (2021). Implementations in machine ethics: A survey. ACM Computing Surveys, 53 (6), Article 132. https://doi.org/10.1145/3419633 van Dijk, T. (1990). La noticia como discurso. Comprensión, estructura y producción de la información [News as discourse: Understanding, structure, and production of information]. Paidós. Wagner, M., Borg, M., & Runeson, P. (2024). Navigating the upcoming European Union AI Act. IEEE Software, 41 (1), 19 - 24. https://doi.org/10.1109/MS.2023.3322913 Walker, G. A. (2024). Artificial intelligence (AI) law, rights & ethics. The International Lawyer, 57 (2), 171 - 306. Watson, D.S., Mökander, J., & Floridi, L. (2024). Competing narratives in AI ethics: A defense of sociotechnical pragmatism. AI & SOCIETY, 40 , 3163 - 3185. https://doi.org/10.1007/s00146 - 024 - 02128 - 2 Weaver, D., McCombs, M., & Shaw, D. L. (2004). Agenda - setting research: Issues, attributes, and influences. In L. L. Kaid (Ed.), Political communication research (pp. 257 - 282). Routledge. Zuboff, S. (2022). Surveillance capitalism or democracy? The death match of institutional orders and the politics of knowledge in our information civilization. Organization Theory, 3 (3). https://doi.org/10.1177/ 26317877221129290 Zylinska, J. (2020). AI art: Machine visions and warped dreams . Open Humanities Press. Meneses - Fernández & Chinea Rodríguez 22 / 22 Online Journal of Communication and Media Technologies , 16 ( 3 ), e202643 APPENDIX A Examples of digital activism and corporate lobbying headlines: 1. Tech companies lash out at new EU data protection measures El Español 21 December 2020 2. IBM criticizes the exclusion of international companies in European projects to strengthen its digital sovereignty El Español 29 May 2022 3. More than 120 organizations ask the Government not to give in to pressure from technology companies on artificial intelligence | The Citizen AI Coalition asks the government to stand firm and not allow large companies to manipulate the regulation of this t echnology for their benefit El Diario 5 October 2023 4. Activist cornering artificial intelligence in the courts: “ It ’ s the fight of our lives ” El País 03 January 2024 5. Meta will not offer its new generative AI models in Europe due to its “ unpredictable regulatory environment ” El País 18 July 2024        