The Role of Generative AI in Enhancing Audience Participation in Journalism: A Scoping Review

Abstract

The emergence of generative AI has led to significant transformations, reshaping media organizations’ workflows and facilitating new forms of participation in the news. This scoping review aims to map the existing literature on the role of generative AI in enhancing audience participation and engagement in journalism. It investigates the theories and methods employed in relevant studies, emerging areas of focus and AI applications used by media organizations to foster audience involvement with the news. The data collection was conducted using online sources (Scopus, EBSCO, ProQuest, and Google Scholar) and a total of 30 studies, published between 2022 and 2024, were selected based on the following criteria: English-language, peer-reviewed, and (hybrid) open-access publications relevant to the research objectives and aligned to the search keywords. The results reveal different theoretical and empirical approaches to studying generative AI in journalism, emphasizing emerging focus areas regarding the adoption of relevant tools and the legal and ethical challenges associated with the credibility of news content. Additionally, generative AI applications in journalism primarily serve to facilitate participation and engagement through personalization, accessibility and interactive storytelling. Lastly, this study highlights the potential of generative AI to enhance audience participation in the news, underscoring the ethical and practical challenges it poses as well as existing research gaps, setting the stage for further research.

1. Introduction

Over the years, journalism has undergone significant transformations due to the technological impact on its fundamental characteristics. From the rise of the World Wide Web (WWW) to the automation of news production through Artificial Intelligence (AI) technologies, the journalist-audience relationship has been reshaped along with the established journalism workflows [1,2,3,4]. Accordingly, readers’ involvement in journalistic practices is considered essential due to the inherently audience-focused nature of news, reaching interest since the early 2000s [5]. In this context, technological evolution has played a pivotal role, changing the way information is produced and consumed, and providing space for audience participation in the news creation [6,7]. Particularly, social media, algorithms, and AI have served as essential technologies, encouraging users’ contributions, automating newsroom workflows, and redefining the dynamics between journalists and the public. Hence, they have created opportunities for two-way communication, ensuring multiple ways for packaging the news, according to readers’ preferences [1,8,9]. Moreover, the ever-evolving field of technology introduces a range of solutions into journalism, transforming its fundamental practices. One such development is the recent evolution of generative AI, which raises new queries about the future of journalism and audience’s role in the news production [10,11].

Since 2022, media organizations worldwide have turned to generative AI, which was expected to change the contemporary media landscape, offering capabilities for optimal news creation through advanced machine learning (ML) systems [10,11,12]. Recent studies provide valuable insights regarding the impact of generative AI on journalism, underlining its potential uses in the news creation, the transformative role of advanced AI chatbots (e.g., ChatGPT) and content generators in the media sector and the ethical concerns about the accuracy of AI-generated content [12,13,14,15]. However, these studies often lack an audience-centered approach, providing limited contributions regarding how generative AI can particularly enhance public’s participation and engagement in journalism.

For these reasons, a scoping review was conducted to systematically map the research on generative AI in journalism, focusing on the theoretical and empirical frameworks used to examine its impact in journalism practice, the recent AI applications that foster audience participation in the news and highlighting potential knowledge gaps. Hence, this paper aims to contribute to the current literature, expanding the analysis of the topics by addressing the following research questions:

• RQ.1: Which theoretical and empirical approaches are currently used to examine generative AI in journalism and its impact on audience participation?
• RQ.2: Which are the target sectors and areas of focus in recent studies on generative AI in journalism?
• RQ.3: How is generative AI used by media organizations to enhance audience participation and engagement with the news?

The findings reveal a combination of different theoretical frameworks and empirical approaches associated with the examination of generative AI in journalism, highlighting emerging areas of focus oriented towards the legal and ethical challenges regarding news reliability, the convergence between disruptive technologies and media operations and the evolving AI-journalism relationship.

2. Theoretical Background

2.1. Audience Participation in Journalism in the Digital Age

In the early 1990s, traditional journalism began to change, as technological evolution led to the emergence of the first news websites [2,3]. Since then, the rise of Web 2.0, digital technologies, and social media helped redefine the established roles between journalists and the public [16]. According to Hermida (2011), technological advances enabled readers to contribute to professional journalism by creating and disseminating news stories on journalistic websites and social media platforms [17]. This shift marked a transition for readers from ‘passive audiences’ to ‘active users’ and eventually to ‘prosumers’, who take on dual roles as both producers and consumers of news. As a result, journalists’ role as ‘gatekeepers’ evolved into that of ‘gatewatchers’, who select, filter and edit readers’ contributions, ensuring news quality and accuracy [5,18,19].

Over the years, many scholars have examined audience’s active involvement in the news-making process, coining terms like ‘citizen journalism’, ‘user-generated content’ (UGC) and ‘participatory journalism’ to describe readers’ contributions to news websites [5]. Among these definitions, ‘participatory journalism’ has been widely used to describe citizens’ active role in collecting, reporting, and disseminating news to provide independent and crucial information, essential for preserving the democratic values [20] (p. 9). According to Saridou and Veglis [21], in participatory journalism journalists and users produce news on a mainstream platform and at the same time content created by users outside the media organization’s platform is acquired and used by professional media outlets. Participatory tools in news media include –among others– citizen and journalists’ blogs, user-submitted content, collective interviews, comment sections, forums, polls, content ranking and social media sharing tools [21,22]. However, research has shown that, although news organizations initially adopted such tools, users had limited access to many of the news production stages while journalists found themselves facing participatory fatigue, trying to prevent and solve problems arising from the integration of UGC in their daily work routines [5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23].

During the late 2010s, news organizations began to incorporate automated technologies into their everyday practices, reshaping the news production processes. This shift marked a new era in the media industry, as algorithms and AI technologies, allowed journalists to quickly and automatically produce news and develop a more interactive relationship with their audience [24,25].

2.2. Redefining Journalism Practices and Audience Participation in News Making Through Automation and Artificial Intelligence

Algorithmic journalism, also known as robot or automated journalism, gained increased interest during the 2010s, introducing a plethora of automated processes that redefined journalistic daily routines [25]. Hamilton and Turner (2009) have argued that algorithmic journalism is based on the notion that algorithms, data, and knowledge from the social sciences enhance the media duty to act responsibly, limiting the spread of inaccurate news [26] (p. 62). In this regard, the role of automation is crucial, representing a multitude of AI-related procedures that include the availability of structured data and algorithms capable of producing inferences. Such developments initially allowed journalists to analyze data and identify patterns and trends from various sources, fostering news automation. Accordingly, semi or fully automated forms of gathering, filtering, composing or sharing the news gradually assumed a greater place in journalism, revolutionizing the production standards. Leading media organizations started using semantic web tools to incorporate UGC into their websites. CNN iReport, Guardian Witness and BBC’s UGC Hub were some pioneering platforms at the time that allowed users to submit news-related content which was intended for publication [8,27,28]. From then on, algorithms were used to automatically produce articles, filter social media content or test new forms of news dissemination while crowdsourced platforms for gathering UGC facilitated data-driven reporting [1,8,29].

Such transformations in journalism workflows provided a framework of experimentation where chatbots, a form of conversational user-interface (CUI), were also used to disseminate news in various platforms, providing users with a greater sense of control over the information they received. In 2016, CNN rolled out a variety of chatbots across messaging apps like LINE, Kik and Facebook Messenger to publish international stories, invite readers to experience the news in a “choose your own adventure” format and allow them to ask and receive responses to open-ended questions. Moreover, ABC’s newsbot could make news available on various platforms, providing readers’ the opportunity to select what information they receive, where and when.

Although journalism’s algorithmic turn was associated with the incorporation of innovative tools into the news production, research has shown that automation, algorithms and AI gradually contributed to the rise of significant challenges, ranging from the cultural fragmentation and ‘echo chambers’ to algorithmic gatekeeping in news dissemination. Algorithms’ influence on the news dissemination processes has favored engagement-driven content over in-depth nuanced reporting, undermining journalists’ credibility. News personalization has also led individuals to consume information that aligns with their interests and beliefs, isolating them in echo chambers that discourage the existence of diverse viewpoints towards a topic [30,31]. According to Newman et al. (2023) [11], personalized algorithms that are currently used to prioritize content from celebrities or influencers over credible news sources, undermine journalists’ control on what is newsworthy, reducing informed public discourse. Lastly, the adoption of generative AI by media organizations has also influenced the news production, creating both concerns and opportunities for attracting new audiences [10,11].

2.3. The Integration of Generative AI Tools in Media Organizations

According to Caswell (2023), the advent of generative AI accompanied with the public release of ChatGPT in late November 2022 has led to global transformations that constantly affect the media industry [10]. Generative AI offers new capabilities, including text generation, natural language conversations (like chatbots), content analysis for generating data (e.g., scores, classifications etc.), and content production (e.g., headlines, summaries or even entire news articles). Additionally, it accomplishes various tasks, following instructions received from external actors by using natural language (e.g., “Write three headlines for the following article: <article text>”) [12]. Furthermore, the incorporation of generative AI tools in news-related procedures is linked with increased possibilities that enhance journalists’ efficiency and productivity [32,33].

Currently, news publishers across the globe are using generative AI tools to improve content production and achieve higher audience reach and engagement, leading readers to pay attention or money for consuming news [34]. Media practitioners in several organizations, such as The Los Angeles Times and the Associated Press, have recently employed similar generative AI technologies, including ML, machine vision, transcription, translation, natural language processing (NLP), natural language generation, social listening, and recommender systems for similar purposes: gathering and processing information, content creation and dissemination, audience engagement and supporting business functions [13,34,35,36].

In this context, the use of generative AI becomes prominent in the news production process. For the phase of news gathering, generative AI can support content aggregation from online sources, trend analysis to identify newsworthy topics but also to develop future scenarios for news events [12]. For news creation, automated article writing and automated transcription, multilingual translation, data analysis and visualization, voice or speech synthesis are some of its common uses. The New York Times is using automated-voice technology to provide audio formats from written articles and translation models to translate news in Spanish [12,37]. In terms of news verification, generative AI enhances real-time fact-checking, verifying claims to reduce the spread of misinformation. The Guardian uses related tools that scan news for inconsistencies, provide journalists correction suggestions, and even alternative phrasing to ensure clarity and accuracy. Moreover, Reuters News Tracer is implementing machine-learning algorithms to detect breaking news from X, looking for clusters of tweets talking about the same event to evaluate their credibility to become news [12,15,38]. In the dissemination and moderation stage, generative AI is used for content personalization, user engagement analysis, social media posting and automated content distribution or search engine optimization. Czech media like Economia and Seznam are using related tools for content moderation, detecting and filtering users’ content from social media and discussion forums to avoid the spread of misinformation [12,32].

The integration of generative AI in strengthening the news production stages can create a more connected and dynamic workflow, focusing on effective information sharing and improving customized content dissemination. By supporting a collaborative and networked model of news creation, it can foster journalists’ productivity, enhance users’ emotional attachments with AI conversational systems and allow them to deeper engage with the news, making it interactive, personalized and reliable [15,39,40]. Through advanced algorithms that analyze user behavior and preferences, generative AI enables the creation of tailored news feeds, headlines and summaries that fit readers’ special interests. Consequently, personalization enhances audience connection with the news. Additionally, it can foster inclusivity, connecting audiences from diverse backgrounds and enriching public discourse with a wide range of perspectives through the presentation of multifaceted views regarding a particular topic. Lastly, while AI and particularly generative AI is associated with the spread of misinformation, it also provides tools that enable journalists to verify news, building a trustworthy media environment where audience feedback loops further enhance accuracy and fostering the adoption of a user-centered approach by media organizations [15,41,42].

3. Materials and Methods

Considering the widespread adoption of generative AI by news organizations since November 2022, this study employs a scoping review to map the existing literature on generative AI impact on journalism and its role in enhancing audience participation in the news. Scoping reviews are methods of knowledge synthesis that identify trends and gaps within a knowledge base to inform research, policy, and practice [43] (p. 2). Scoping reviews can also map an existing body of literature based on factors such as time, location (e.g., country or context), source (e.g., peer-reviewed or grey literature), and origin [44] (p. 142). The methodology follows Arksey and O’Malley’s [45] stages for scoping reviews, alongside Engelke’s [46] codebook for systematically analyzing, categorizing and interpreting studies on participatory journalism. Moreover, the Preferred Reporting Items for Systematic Reviews and Meta-Analyzes extension for Scoping Reviews (PRISMA-ScR: https://www.prisma-statement.org/scoping (accessed on 24 August 2025)—a checklist of 22 reporting items, from which the optional ones (12 and 16) as well as those related to funding (22) were not applied—was also adopted as protocol for guiding the conduct and reporting of this review. Through these frameworks, scholarly works on generative AI in journalism and its impact on audience participation were collected, examined, categorized, and interpreted to serve this review’s purpose.

Particularly, the first stage of Arksey and O’Malley’s (2005) framework for scoping reviews involved formulating research questions while the second one focused on identifying relevant scholarly works by searching across different sources. It also included decisions on sources, search terms, language, and publication year(s) to ensure their relevance with the study objectives [45]. Accordingly, Scopus, EBSCO, ProQuest, and Google Scholar were selected to search for relevant works. These sources were chosen as they offer access to multiple databases and indexed, peer-reviewed journals in communication and journalism, facilitating the identification of credible publications relevant to the study’s aims. In addition, filters, such as language (English), full-text scholarly works, peer-reviewed journals, publication year and database selection, were applied to the respective sources. In Google Scholar the results were manually screened, using the same parameters as this database only allows filtering by publication year. All filters aimed at identifying scholarly works relevant to the research inquiries and compatible with the study’s selection criteria analyzed in the third stage.

It is worth noting that, unlike Scopus, EBSCO, and ProQuest required selecting specific databases for the search to be conducted. Thus, the Library, Information Science and Technology Abstracts (EBSCO) alongside ProQuest Central and ProQuest Publicly Available Content were chosen as they were thematically aligned with the aims of the study, including peer-reviewed journals pertinent to the field of communication, thereby yielding more relevant results in line with the selection criteria.

With respect to the publication years, the period from 2018 to 2024 was initially selected to capture studies that may have involved pilot testing and experimental uses of AI-related technologies. However, the results encompassed publications from unrelated fields (e.g., health and computational sciences), while communication studies, particularly between 2018 and 2022, focused on issues such as social media use by content creators and online strategies of political parties during elections. In contrast, the period from 2022 to 2024 yielded publications that both align with the study’s objectives and document experimental applications of generative AI technologies by media outlets. Therefore, the search was finally conducted between 2022 and 2024, the years followed the widespread use of such technologies by the media sector, using the following query string: (“generative AI” OR “AI-generated content”) AND (“journalism” OR “news” OR “news content”) AND (“audience participation” OR “audience engagement” OR “reader interaction”) AND (“open access”). The term open access was used to identify journals that met the inclusion criteria, especially in case of Google Scholar, while in Scopus was replaced by respective filters.

During the third stage, studies were selected based on specific inclusion and exclusion criteria, similar to systematic reviews, which were designed to address the research questions, facilitating the presentation of insights relevant to the study’s purposes [45]. The inclusion criteria are provided in

Table 1. Inclusion criteria.

Criterion	Application in Study Selection
English-language publications	Accurate data examination, as authors are proficient in English, ensuring the presentation of comprehensible results
Peer-reviewed journals	Expert evaluation, ensuring accuracy and credibility of each scholarly work, similar to Engelke’s corresponding criterion [46]
Open-access publications, open access and hybrid open access publishing model	All types of open-access publications (e.g., research articles, commentaries, review papers etc.) included in journals with both open access and hybrid open access publishing model, enriching the results with diverse contributions and supporting transparency with publicly accessible information
Keyword relevance	Publications whose titles, abstracts, and keywords matched the search terms
Topic relevance	Publications whose topics were relevant to the review’s aims, providing insights into how media outlets are using generative AI in enhancing audience participation and engagement with the news

.

Specifically, of the 736 studies initially retrieved, 200 were identified as duplicates, appearing in more than one database, and were excluded. The process continued with the remaining studies (536), which were examined by the authors according to the review’s inclusion and exclusion criteria. At this point, 300 studies were excluded as they were linked to irrelevant research topics that emphasized the relation between journalism and innovative technologies rather than the role of generative AI in enhancing audience participation and engagement in the news, after manually examining their abstracts. The remaining 236 studies were subsequently assessed in detail by the authors, considering titles, abstracts, keywords and the full texts. Of these, a total of 206 were excluded based on the following exclusion criteria: (1) non-open access publications (including paywalled studies) to ensure publicly accessible data, (2) foreign-language publications (with English abstracts but full texts in different languages) to maintain accuracy and clarity in the presentation of results, and (3) the grey literature (non-peer-reviewed studies) to avoid the presentation of data lacking expert evaluation and to prevent inconsistencies with the inclusion criteria. To achieve an accurate and consistent review, two of the authors independently extracted the data from all eligible studies. The extracted data were reviewed by the third author. If the legibility of an article was in doubt, a discussion among the three authors was conducted to reach a consensus. Through this process, 30 studies were identified as relevant and were included in the review (Figure 1).

The fourth stage involved charting, which syntheses qualitative data by organizing it around key themes, akin to the data extraction process in systematic reviews [45]. As mentioned above, this study employs Engelke’s (2019) codebook, which was developed through a systematic review on online participatory journalism, using specific variables to analyze and organize the respective data [46]. In this vein, the theoretical approach focused on examining the theoretical contexts used by the authors of the selected publications to contextualize and examine generative AI impact on journalism. The empirical approach aimed to identify and categorize (via structured information—tables) the research methodologies, providing insights on the geographic distribution, the target sectors and focus areas of the respective works. At the same time, the results provide insights regarding the adoption of generative AI in journalism for enhancing audience participation and engagement with the news. Similarly to Engelke’s (2019) methodology which considers the “results regarding journalist–audience relationship” as an openly coded qualitative variable, it was adapted to capture evidence on generative AI technologies and tools that enhance audience participation and engagement with the news [46] (p. 35). The fifth stage included collating, synthesizing and outlining the results, providing a general overview of all material reviewed, followed by an extended analysis of each work’s content. The sixth stage was additional and was not included as part of this study’s methodology [45].

4. Results

The total number of publications analyzed is 30 (n = 30), from the total of 736 results returned by EBSCO, ProQuest, Scopus, and Google Scholar (The full list of scholarly works which were analyzed can be found in Supplementary Materials, following the main text). Most of them are articles (23), while colloquiums/commentaries (2), reports (1), reviews (1), experimental studies (1), technical (1), and position papers (1) are relatively rare. Moreover, the open-access publishing model is more common than the hybrid open-access, including 10 out of 15 journals recorded in the study. Subsequently, 25 out of the 30 scholarly works have been published in peer-reviewed open-access journals, with Emerging Media being the dominant (seven publications), followed by Journalism and Media (five publications), Profesional de la Información (four publications), AI Magazine (three publications), Humanities & Social Sciences Communications, Review of Communication Research (RCR), Arab Media & Society, BiD, International Journal of Maritime Engineering (IJME) and Digital Business, each with one publication. On the contrary, five scholarly works have been published in hybrid open-access journals, including Information & Communications Technology Law (one publication), AI & Society (one publication), Journal of Media Business Studies (one publication), Journalism & Mass Communication Educator (one publication), and International Journal of Information Management (one publication).

Furthermore, nine out of the 15 journals adopted an interdisciplinary approach, combining journalism with humanities, social and behavioral sciences, computer science, engineering, information management, and digital technologies. These journals include AI & Society, Information & Communications Technology Law, Humanities & Social Sciences Communications, Digital Business, AI Magazine, International Journal of Information Management, International Journal of Maritime Engineering (IJME), Profesional de la Información, and BiD. On the contrary, the remaining ones are communication journals that focus on journalism and media studies (disciplinary approach).

Regarding the publishers, Sage publishes Emerging Media and Journalism & Mass Communication Educator, including eight out of the 30 scholarly works. Subsequently, MDPI publishes Journalism and Media (five publications), Oxbridge Publishing House publishes Profesional de la Información (four publications), the Association for the Advancement of Artificial Intelligence (AAAI) publishes AI Magazine (three publications), Elsevier publishes International Journal of Information Management and Digital Business (one publication per journal) and Springer publishes Journal of Media Business Studies and Humanities & Social Sciences Communications (one publication per journal). Routledge, Taylor & Francis, Adham Center for Television and Digital Journalism (American University in Cairo), The Royal Institution of Naval Architects and the Faculty of Information and Audiovisual Media (University of Barcelona) each publish one journal with one publication: Journal of Media Business Studies, Information & Communications Technology Law, Arab Media & Society, International Journal of Maritime Engineering (IJME), and BiD, respectively. The Review of Communication Research (RCR) is a self-published journal that also includes one publication. Lastly, from the total of 30 studies included in the review, 25 were published in 2024, four in 2023 and one in 2022 (

Table 2. General overview of publications and their journal attributes.

Number of Publications	n = 30
Type of Publication
Articles	23
Colloquiums/Commentaries	2
Reports	1
Reviews	1
Experimental Studies	1
Technical Papers	1
Position Papers	1
Journal Publishing Model
Open-access	10
Hybrid open-access	5
Type of Approach
Interdisciplinary approach	9
Disciplinary approach	6
Publisher
Sage	2
MDPI	1
Routledge	1
Taylor & Francis	1
Springer	2
Oxbridge Publishing House	1
Association for the Advancement of Artificial Intelligence (AAAI)	1
Elsevier	2
Adham Center for Television and Digital Journalism—American University in Cairo	1
The Royal Institution of Naval Architects	1
Faculty of Information and Audiovisual Media—University of Barcelona	1
Review of Communication Research (RCR)	1
Publication Year
2024	25
2023	4
2022	1

).

4.1. Theoretical and Empirical Approaches to AI and Generative AI in Journalism

Research question (RQ.1) deals with the theoretical and empirical approaches used by the reviewed studies to examine generative AI in journalism and its impact on audience participation. The theoretical approaches identified include the cultural political economy approach, media framing, the institutional theory in conjunction with digital inequality concepts, the technological appropriation theory accompanied with the socio-technical concept of technology adoption, AI-media concept along with the concept of mediatization and the theory of disruptive innovation. Moreover, the uses and gratifications (U&G) theory, innovation theory, the framework of studying media participation in the public debate, Porter’s value chain, activity theory (AT), labor theory, audience perception criteria for data-driven journalism, and the unified theory of acceptance and use of technology (UTAUT) have also been detected within the examined studies. These frameworks focus on the impact of AI and generative AI on journalism alongside the factors influencing audience engagement with machine-generated content, providing a broader view of these areas.

More specifically, the cultural political economy approach was applied to explain the influence of dominant political and economic forces in legitimizing the discourse surrounding AI-generated news. Through these lenses, generative AI is presented to be shaped by economic exploitation and power imbalances, benefiting large AI companies, limiting the diversity of voices in journalism and posing challenges that undermine democratic values, such as objectivity and diversity.

Through media framing the adoption of AI in journalism is depicted as a mix of both risks and opportunities, including job displacement, ethical concerns and potential threats to editorial independence.

The institutional theory, accompanied by digital inequality concepts, emphasizes how both AI and generative AI adoption are shaped by established power dynamics within media landscapes. Thus, research highlights how the dominance of Western media and the respective power asymmetries lead to unequal distribution and access to AI resources, enhancing disparities across news organizations.

The technological appropriation theory, along with the socio-technical concept of technology adoption, interprets the socio-technological barriers to the adoption of AI and generative AI in newsrooms in weak economies.

AI-media, along with the concept of mediatization, argues that AI is reshaping contemporary media practices, offering a more holistic view of its impact on the field of journalism. Both approaches provide a comprehensive and structured framework for understanding the convergence between AI technologies and media operations, emphasizing their role in media, particularly in the context of emerging technologies such as generative AI.

The theory of disruptive innovation assists in understanding how journalism institutions adapt to disruptive technologies like AI and generative AI, highlighting the integration of such technologies into journalism and interpreting the challenges institutions face in adapting to these changes while maintaining their relevance.

The uses and gratifications (U&G) theory was applied to determine the factors contribute to an increased level of consumer engagement with videos generated by using ChatGPT, underlying the innovative nature of such content as significant trait that attracts new audiences.

Innovation theory focuses on how journalists comprehend and respond to the constant transformations within the media landscape, interpreting their perceptions in terms of using cutting edge technologies and indicating both the ethical and social challenges linked to the adoption of AI by contemporary newsrooms.

The Porter’s value chain framework suggests that primary and secondary activities within media organizations can create value, enhancing the understanding of the activities where local newsrooms see the greatest potential for AI to deliver benefits and highlighting the consequences of related technologies in news quality and accuracy.

Activity theory (AT) can be used to identify key components in misinformation generation and mitigation, providing a comprehensive overview of the mechanisms linked to crisis misinformation to support decision-making processes and efforts that minimize possible harm in such scenarios.

The labor theory examines relations of production in industrial capitalism, contending that technology, including AI and generative AI, and financial influence have reshaped journalists’ roles, disorienting the tasks they perform and transforming the news production process.

Audience perception criteria for data-driven journalism cover antecedents of perception, emotional and cognitive impacts, article composition, and news/editorial values to provide an understanding of how audiences engage with and evaluate AI-enhanced journalism.

The unified theory of acceptance and use of technology (UTAUT) examines the factors influencing people’s decisions to adopt and use new technologies, interpreting audience responses to AI-generated versus journalist-written content.

Most of the theoretical approaches are primarily concerned with the role of generative AI in reshaping contemporary journalism. These frameworks offer diverse perspectives on how generative AI is influenced by both the economic and social structures in which it is applied, highlighting the complexity between technological, political, and institutional factors that determine contemporary media landscapes and indicating audience perceptions regarding the consumption of such content alongside the opportunities arise in the news production. On the contrary, media framing, innovation theory and Porter’s value chain provide a broader context of discussion, emphasizing media’s role in forming public attitudes regarding the acceptance (or not) of AI solutions, defining areas where its application can increase profit and underlying both the ethical and social challenges associated with its adoption by news outlets.

Regarding the empirical approaches and particularly the research methodology, single-method studies (23) are more common than multi-method (7). Among the single method studies, qualitative research appears to be more common than quantitative research. Within qualitative methods, literature reviews and interviews are dominant, followed by case studies and content analysis, while focus groups, archival research and benchmarking analysis are less commonly used. Within quantitative methods, experiments and numerical data analysis appeared in the same number of studies. Multi-method studies typically employ mixed-method designs, including mixed data analysis, convergent parallel design (CPD), exploratory sequential design (ESD) systematic review and bibliometric and content analysis, offering an overview of both numerical and non-numerical data composing the findings (

Table 3. Research methodology.

Research Design	Method Type	Number of Publications
Single-method studies	Literature reviews	6
	Interviews	6
	Case study	2
	Content analysis	2
	Benchmarking analysis	1
	Archival research	1
	Focus groups	1
	Numerical data analysis	2
	Experiments	2
Multi-method studies	Mixed data analysis	2
	Convergent parallel design (CPD)	2
	Exploratory sequential design (ESD)	1
	Systematic review	1
	Bibliometric and content analysis	1

).

Furthermore, 18 out of the 30 scholarly works are comparative in nature, whereas a small number (8) focus on a global level, covering North and Latin America, the United States, EU countries, Asia and Africa. Additionally, studies concentrating on western regions, particularly EU countries like Germany, Belgium, Denmark, Spain, Finland, France, United Kingdom (former EU member), Netherlands, Ireland, Italy and Sweden, and the United States are prevalent (13) while those focusing on non-Western regions are less frequent (7). These studies emphasize African countries, such as South Africa, Nigeria, Kenya, Ghana, Zimbabwe, Botswana and Eswatini, the Arab region, including Egypt, Quatar, Saudi Arabia, Cairo and Kuwait, and India. Lastly, few of the examined studies do not focus on a specific geographical area (2).

4.2. Target Sectors and Areas of Focus on Generative AI in Journalism

Research question (RQ.2) concerns both the target sectors and areas of focus of the recent studies on generative AI in journalism. In terms of the target sectors, single-method studies mostly focus on media organizations (16), followed by journalistic articles (2) while few of them emphasize media laboratories (1), AI tools for creative content creation (1), AI-generated content by Youtubers and journalists (1), cross-platform content (1) and AI companies and news outlets (1). In terms of focus areas, the largest number of studies concentrate on the adoption and impact of generative AI on journalistic practices (10), followed by the ethical and legal challenges of generative AI in journalism (4), the interplay between generative AI and news content generation (3), the evolving AI-journalism relationship (3), the convergence between AI technologies and media operations (2), and the impact of disruptive technologies, including generative AI, on journalism education (1). Mixed-method studies concentrate on media organizations (2), journalistic articles (1), scholarly works (1), AI companies and news outlets (1), and emerging research topics and trends related to generative AI in the field of journalism (2). The dominant areas of focus include the adoption and impact of generative AI on journalistic practices (3), followed by the established AI guidelines across media organizations (1), the convergence between AI technologies and media operations (2) and the evolving AI-journalism relationship (1) (

Table 4. Target sectors in single and multi-method studies.

Study Type	Target Sector
Single method	Media organizations
	Journalistic articles
	Media laboratories
	AI tools for creative content creation
	AI-generated content by YouTubers and journalists
	Cross-platform content
	AI companies and news outlets
Multi method	Media organizations
	AI companies and news outlets
	Journalistic articles
	Scholarly works
	Topics and trends in generative AI in journalism

and

Table 5. Areas of focus in single and multi-method studies.

Study Type	Areas of Focus
Single method	Adoption and impact of (generative) AI on journalistic practices
	Ethical and legal challenges of generative AI in journalism
	Interplay between generative AI and news content generation
	Convergence between AI technologies and media operations
	Evolving AI-journalism relationship
	Impact of disruptive technologies on journalism education
Multi method	Adoption and impact of (generative) AI on journalistic practices
	Established AI guidelines across media organizations
	Convergence between AI technologies and media operations
	Evolving AI-journalism relationship

).

4.3. Use of Generative AI by Media Organizations for Enhancing Audience Participation and Engagement

Research question (RQ.3) concerns the use of generative AI by media organizations for enhancing audience participation and engagement with news-related content. Although all examined studies (n = 30) argue that generative AI has the potential to ensure audience’s active role in news-related processes, 17 provide specific insights to describe how this is achieved in contemporary media organizations:

• Experimental uses and working examples of news products and publishing workflows similar to LLMs and generative AI

Several news organizations had developed working examples of news products and publishing workflows similar to generative AI and LLMs prior to ChatGPT’s public release, which were structured and automated journalism projects. For instance, BBC’s interactive and chat interfaces to news are early conversational experiences similar to those provided by LLMs. These interfaces allowed audiences to ask for additional information about a news story and they were published in two certain forms: as text chat bots and as an interactive news experience on the Amazon Alexa voice agent known internally at the BBC as Skippy. Moreover, the Explainer Builder project was developed to provide explainers for readers who were less familiar with a news topic but also to prevent frequent news consumers from being overloaded. This initiative had limited uses because the burden of producing and maintaining explainers by hand was too great. However, it paved the way for the development of “Tell Me More” project that leverages LLMs to automatically generate draft explainers.

Furthermore, Bloomberg has recently incorporated generative AI and LLMs into the News Innovation Lab’s toolkit to explore how they can generate solutions and applications that better serve the readers. European public broadcasters like RAI (Italy) and NPO (Netherlands) have also explored these technologies to align their services with the evolving standards of the audiovisual ecosystem. RAI tested the capabilities of natural language processing (NLP), computer vision, and visual information analysis in the news production, focusing on enhancing news personalization and content recommendation systems. NPO carried out experiments on the use of related technologies to generate sports news and election results from structured data.

Similarly, Indian and digital-native outlets like Rappler (Philippines) and The Daily Maverick (South Africa) have tested AI tools for content personalization and the creation of engaging visual stories while US local media point out the potential of generative AI to create value and enhance journalistic tasks like automated writing and news reporting.

• AI tools for video automation services

AI tools are also used to create multimedia content that captures audience attention. For example, news organizations such as BBC, Reuters, and The Economist have adopted video automation services from providers like Wibbitz, Wochit, and Synthesia, which allow for fast, visually appealing video production tailored to audience interests.

• AI-generated news anchors and robot news readers

Newsrooms in economically developed countries (e.g., Sharjah Media City in the United Arab Emirates) have adopted AI-generated virtual news anchors to address problems such as the late arrival of news presenters, and to distribute timely online news bulletins (e.g., Kuwait News has introduced “Fedha”, an AI-generated female news anchor). Moreover, AI-script-to-voice formats have been adopted to improve accessibility for disabled individuals (e.g., AI Ain News in Abu Dhabi, UAE). These innovations aim to enhance audience participation and engagement by offering accessible, consistent and inclusive news delivery, addressing viewers’ special needs and increasing the reach and appeal of news content in an increasingly digital and inclusive digital media environment.

Furthermore, broadcasting newsrooms, especially those in African regions, have introduced AI-driven robot news readers to enhance news delivery, by automating the reading of scripts generated by professional journalists or other AI tools to provide real time news updates with natural tone and accurate pronunciation. AI robot news readers can broadcast in multiple languages, including local dialects, to ensure that news content is accessible to diverse audiences. They also enable constant 24/7 news coverage, providing media organizations with the opportunity to stay on top of breaking news without relying on human presenters. As such, AI-news readers tailor journalistic content to specific regional preferences, enhancing audience engagement and ensuring that news is both timely and relevant.

• AI-driven systems for news recommendations and content personalization

News organizations tend to adopt AI-driven systems for news recommendations and content personalization, delivering tailored content that matches individual preferences and behaviors. These systems detect and analyze user data, such as past interactions within social media and news websites, interests or engagement patterns, with the aim of creating personalized news feeds. In this context, technology companies offer media outlets like El País, La Razón, RTVE and El Español (all based in Spain) AI tools like Lynguo, Tailorcast and OTTforyou that include services such as audience segmentation analysis, podcast recommenders, social network and competitor trend analysis, enhancing engagement by providing readers’ content tailored to their interests.

• Transcription robots for real time news

As happens with AI script-to-voice formats, news organizations have introduced transcription robots that enable real-time transcription of live radio and TV shows in indigenous languages, which are streamed online. Real-time transcription tools enhance audience engagement and participation in the news, making content more accessible, especially to individuals who speak indigenous languages, fostering greater inclusion. By transcribing content in real time, these tools allow audiences to engage with live broadcasts even if they miss or are not able to watch the original ones. Therefore, the transcription process improves content relevance and availability, leading to higher audience interaction and participation in news events, discussions, and debates.

• AI-tools for interactive and dynamic storytelling

Contemporary newsrooms, especially those in non-western countries, have developed and/or introduced AI-driven text-editing tools that enhance interactive storytelling, fostering more dynamic and engaging news narratives that attract new audiences. By using generative AI technology to create multimedia forms of content, news media can increase readers’ interactions with the news.

• AI-tools for immersive news experiences

The use of AI-tools in immersive journalism, including both VR and AR experiences, has emerged as an innovative way to create more empathetic news experiences, enhancing audience understanding and engagement with complex narratives. Although such journalistic initiatives are quite rare so far, the recent example of BBC’s “Damming the Nile” VR experience highlights the transformative potential of such tools in revolutionizing the way journalistic content is produced and consumed.

• AI-driven meta data tagging

News organizations have currently used generative AI technology to automate meta-data tagging, by analyzing news content or by generating descriptive tags (e.g., topics, keywords etc.) to improve news classification, discoverability and personalization. Thus, users are more capable of navigating, discovering and engaging with journalistic content that aligns with their preferences and special needs.

• Generative-AI in journalism education

The integration of generative-AI tools, like ChatGPT, into journalism education, including the respective institutions, reflects an increasing awareness of AI’s potential to transform the existing relationships and alter journalism practices. Such transformations revolutionize how media organizations produce and distribute news content, transforming journalist-audience interactions, offering a wide variety of tool and related practices, which can foster participation in the news production procedures and develop a more open information environment.

5. Discussion

The aim of the present study was to provide a deeper understanding of the evolving landscape of generative AI in journalism, particularly regarding its potential to reshape audience participation. For this reason, a scoping review focusing on the last two years was applied to explore the existing literature, identify trends and gaps and gain specific insights in the field. The analysis of the theoretical approaches used by the scholarly works studied reveals diverse perspectives on the adoption and implications of AI and generative AI in journalism, highlighting both the opportunities and challenges these technologies present. Specifically, most theoretical approaches applied emphasize the multifaceted and often conflicting dynamics accompanying the integration of AI technologies in the media landscape. Economic power structures and established media hierarchies are suggested to reinforce the challenges posed to media organizations, not only in weaker economies. The cultural political economy, the institutional and the technological appropriation theory suggest that limited access to resources may impede diversity and equality in adopting generative AI tools. Such obstacles are steadily present in the discussion about integrating technology into newsrooms [5,47,48] and generative AI adoption does not seem to be an exception. Moreover, the balance between technological adoption and maintenance of foundational principles and societal roles in journalism arises as a core prerequisite in relevant scholarship through time [6,49]. In the current study, approaches such as media framing and the AI-media concept acknowledge the potential benefits of generative AI integration while considering the ethical and practical risks involved. These perspectives reflect the tension between innovation and disruption, where AI technologies may enhance journalistic practices while simultaneously threatening editorial independence and job security.

As far as the research methodology is concerned, most works followed qualitative research, particularly literature reviews and interviews, suggesting an interest in understanding the subjective experiences and perceptions of AI adoption in journalism. However, the limited use of mixed and quantitative methods may lead to a gap in large-scale, data-driven studies that could offer more analytical and comprehensive insights in the future. Additionally, the global focus of comparative studies underscores an attempt to capture diverse perspectives, while research exclusively on non-western media contexts seems to be less prevalent in the scholarly works studied. Such findings highlight the need towards the emerging journalism exploration that investigates the field from a non-western perspective, responding to the growing calls for reconceptualization of the field towards a broader research framework [50].

Regarding the adoption of generative AI for enhancing audience participation and engagement in journalism, the present study highlights several key aspects that can shed light on the evolution of the relevant existing literature in the field. The analysis reveals that generative AI applications in journalism primarily serve to enhance audience engagement through personalization, accessibility and interactive storytelling. In line with previous technological shifts that were believed to place audiences at the center of news consumption [51,52], both generative AI tools offer capabilities for tailoring content to individual preferences while simultaneously enabling broader accessibility, for example, through multilingual transcription and AI-generated news anchors. It is thus worth noting that such practices pose significant challenges to traditional journalistic foundations, since they can potentially redefine the journalist-audience relationship by introducing non-human intermediaries. This shift echoes earlier transformations from ‘gatekeeping’ to ‘gatewatching’ [18] but takes it further by potentially eliminating human intervention altogether in certain aspects of the news presentation. However, relevant research which is constantly growing [53,54,55] underlines the need for attention to issues of credibility, accuracy and audience acceptance. Specifically, previous findings suggest that while generative AI models and tools, such as AI anchors, improve efficiency and allow for both engaging news and tailored content delivery, they encounter challenges in replicating the emotional resonance of human hosts and in ensuring journalistic trustworthiness, thereby influencing audience attitudes and cultural contexts [53,54,55].

6. Limitations and Future Research

The present review is subject to several limitations that should be acknowledged. First, it does not include grey literature, such as reports or white papers, which may have excluded valuable insights from different sources. Moreover, no formal quality assessment of the included studies was conducted. Following suggestion by Arksey and O’Malley (2005), a scoping review does not aim to assess the quality of evidence and consequently cannot determine whether the findings of studies are generalizable [45]. Another limitation is the focus on open-access publications. While ensuring accessibility, it may have introduced bias by excluding relevant research published in subscription-based journals, thereby narrowing the scope of the findings.

Finally, the present scoping review indicates a notable gap in research regarding audience perspectives on generative AI adoption in journalism. While numerous studies examine how news organizations implement AI technologies to enhance audience engagement, few directly investigate how audiences perceive and respond to AI-generated content. This gap carries the risk of limiting the understanding of whether generative AI truly fulfills its promise of fostering meaningful audience participation. In recent years, scholarly interest in audience perspectives on participatory journalism has grown [56,57], yet there is space for in-depth empirical research directly engaging with users. At this point, future research could elaborate on exploring the lived experiences, perceptions, and attitudes of media users through interviews or other qualitative methods. Taking also into account the issues of trust and credibility, the focus on users’ perspective would provide a more comprehensive understanding of the ever-evolving relationship between audience and the adoption of technological innovations in journalism.

While generative AI offers promising tools for enhancing audience participation in journalism, its implementation requires careful consideration of ethical, social, and political implications. As news organizations continue to navigate this technological frontier, balancing innovation with journalistic integrity remains a critical challenge. To this direction, the establishment of specific editorial guidelines and ethical standards for the use of generative AI in each newsroom would ensure that innovation does not compromise accuracy, transparency, or leads to violation of intellectual rights. Moreover, the development of training sessions for journalists to critically engage with AI tools would allow the exploitation of such tools without downgrading the dialogue with audience members. Overall, a balanced approach between efficiency gains and protection of core journalistic values seems more than essential for the profession.