Project – Trust Calibration in Artificial Intelligence-Generated News: The Influence of Source Disclosure on News Credibility Among Digital Media Users in Lagos State.

CHAPTER ONE

INTRODUCTION

Research Topic: Trust Calibration in Artificial Intelligence-Generated News: The Influence of Source Disclosure on News Credibility Among Digital Media Users in Lagos State

1.1 Background to the Study

The advancement of artificial intelligence (AI) has introduced significant changes in the production, processing, presentation and distribution of news content. AI technologies, particularly generative AI systems, can produce news summaries, headlines, reports, explanatory articles and other forms of journalistic content with limited or substantial human involvement. These developments have created opportunities for faster information processing, expanded content production and improved newsroom efficiency. However, they have also raised questions about accuracy, accountability, transparency and public confidence in news. Unlike conventional journalism, in which readers generally expect journalists and identifiable media organisations to take responsibility for published information, AI-generated news may involve complex relationships between automated systems, human editors, data sources and publishing organisations. This makes it important for audiences to understand not only what a news report says but also how it was produced and who is responsible for its accuracy. Clerwall (2014) examined audience perceptions of automated journalism and demonstrated the importance of understanding how readers evaluate computer-generated news in comparison with traditionally written news. Similarly, Tandoc, Lim and Ling (2020) discussed the opportunities and challenges associated with journalism in the age of automation and artificial intelligence, highlighting the need to consider the consequences of technological change for journalistic practice. These developments provide the foundation for examining how disclosure of AI involvement may influence the credibility that digital media users assign to news content.

News credibility is a central concept in journalism and communication because it concerns the extent to which audiences perceive news content, its sources and the organisations publishing it as believable, accurate, fair and trustworthy. News consumers frequently make judgments about credibility without independently verifying every claim contained in a report. Such judgments may depend on the perceived reputation of the media organisation, the clarity and consistency of the information, the presentation of evidence, the perceived competence of the source and the perceived absence of misleading intentions. Appelman and Sundar (2016) developed a message credibility scale that emphasises perceptions of accuracy, authenticity and believability, providing a useful basis for understanding how audiences assess communication content. Metzger and Flanagin (2013) also explained that people increasingly rely on various credibility cues and information-evaluation strategies when navigating digital information environments. These insights are relevant to AI-generated news because audiences may use the identity of the content producer as an additional cue when judging whether a story deserves belief. A news report labelled as AI-generated may be evaluated differently from an otherwise similar report attributed to a human journalist, even where the content appears comparable. Consequently, the integration of AI into news production raises an important question about whether audiences judge a report primarily by its content, its source, its production process or a combination of these factors.

The concept of trust calibration provides an important perspective for understanding these credibility judgments. Trust calibration refers to the alignment between the level of trust a person places in a system and the system’s actual capabilities, limitations and reliability. In the context of AI-generated news, the objective should not be to persuade audiences to trust every AI-generated report or to encourage them to reject all content produced with AI. Rather, the goal is to help audiences develop an appropriate level of confidence based on the quality of the evidence, the nature of AI involvement, the extent of human verification and the accountability of the publishing organisation. Lee and See (2004), in their influential work on trust in automation, argued that appropriate reliance on automated systems depends on the relationship between a system’s capabilities and the user’s understanding of its performance. Although their work was not restricted to journalism, its principles are relevant to news consumption in an environment where automated systems increasingly contribute to information production. Applied to AI-generated news, trust calibration involves enabling users to distinguish between circumstances in which AI has merely assisted a journalist and circumstances in which AI has generated substantial portions of a report. It also involves recognising that disclosure alone does not establish the accuracy of a story. The present study therefore considers source disclosure as a potential mechanism through which digital media users form judgments about AI-generated news and determine how much confidence to place in it.

Source disclosure is the communication of relevant information about the origin, authorship, production process or technological assistance involved in creating a piece of content. In AI-generated news, disclosure may take several forms, including a label stating that a report was generated by AI, a note indicating that AI assisted a human journalist, an explanation that an editor reviewed the content, or information identifying the organisation responsible for checking the report. These disclosures are not necessarily equivalent because they communicate different levels of machine involvement and human accountability. A statement that a report was “generated by AI” may create a different impression from one indicating that it was “prepared with AI assistance and verified by a journalist.” The wording, visibility, placement and specificity of a disclosure may therefore influence the interpretation of a news report. Toff and Simon (2025), in their experimental study of AI disclosure and audience trust in news, found that labelling journalistic content as AI-generated could reduce perceived trustworthiness, although disclosure of the sources used in producing the content helped counteract this negative response. Their findings suggest that disclosure is not simply a matter of informing audiences that AI was involved; it can also shape their understanding of the process used to establish the report’s credibility. Consequently, investigating source disclosure requires attention to the information communicated by the disclosure and the way audiences interpret it.

The relationship between AI disclosure and news credibility is complex because transparency can produce both reassuring and sceptical responses. On one hand, informing audiences about AI involvement may demonstrate openness, reduce uncertainty and allow readers to make more informed decisions about whether to rely on a report. On the other hand, the same disclosure may cause readers to question whether the report was adequately verified, whether the system introduced errors or whether the media organisation has transferred too much editorial responsibility to automated technology. Morosoli, van der Goot, Resendez, de Vreese and Helberger (2025) examined the effects of AI transparency labels on credibility perceptions and engagement across news topics. Their experimental findings indicated that AI attribution could reduce perceived source credibility and negatively affect sharing intentions in political-news contexts, illustrating that the effects of disclosure may depend on the subject matter and the outcome being measured. This suggests that a disclosure label cannot automatically be assumed to increase credibility simply because it improves transparency. Instead, audiences may interpret disclosure through their existing beliefs about AI, their expectations of journalism and their perceptions of the sensitivity of a particular news issue. The distinction is particularly important for a study of trust calibration because appropriate trust depends not only on whether users know AI was involved but also on whether the information supplied helps them make a balanced judgment about the report.

Digital media have made these questions increasingly relevant by providing multiple channels through which people encounter, evaluate and share news. News may be accessed through online newspapers, digital television platforms, news websites, social networking services, messaging applications and other internet-based channels. In these environments, a user may encounter a headline or short news summary outside the original publisher’s website, where contextual information about authorship, editorial review and source verification may be limited. The rapid circulation of content can also make it difficult for users to distinguish between professionally produced news, AI-assisted journalism, fully AI-generated material and misleading content presented in a news-like format. Metzger and Flanagin (2013) explained that credibility assessment in digital environments involves a combination of source-related cues, message characteristics and users’ information-evaluation practices. Appelman and Sundar (2016) similarly demonstrated the value of measuring message credibility through identifiable dimensions rather than treating credibility as an undifferentiated judgment. These perspectives indicate that the credibility of AI-generated news should be examined as a perception formed through users’ interaction with both the message and the information provided about its production. They also support investigating whether source disclosure helps digital media users distinguish between different levels of AI involvement and assess news content more carefully.

In Nigeria, and particularly in Lagos State, the subject is relevant to the relationship between technological change, digital news consumption and public confidence in information. Lagos is an important setting for studying digital media users because it contains a diverse population of students, professionals, business operators, public-sector workers, journalists and other individuals who may encounter news through different digital channels. These groups may differ in their familiarity with AI tools, their expectations of media organisations and their willingness to accept information attributed to automated systems. The study does not assume that all digital media users in Lagos have the same level of AI awareness or that they respond uniformly to disclosure labels. Rather, it recognises that credibility judgments may vary according to individual experiences, news-consumption habits and perceptions of the organisations publishing the information. The wider literature on automated journalism and digital credibility provides a basis for investigating these differences, while the work of Tandoc et al. (2020) underscores the need to examine how the incorporation of AI affects journalistic practice and the relationship between news producers and audiences. Examining digital media users in Lagos State can therefore provide context-specific evidence about how source disclosure is understood and how it relates to perceived news credibility in a Nigerian urban media environment.

Despite the growing attention given to AI in journalism, an important issue remains: the presence of a disclosure label does not necessarily guarantee that audiences will develop appropriately calibrated trust. Users may interpret disclosure as evidence of transparency, as a warning about possible inaccuracy, or simply as a technical detail that does not affect their judgment. Furthermore, a general AI label may not clearly distinguish between content generated entirely by a machine and content produced with AI assistance under human editorial supervision. Toff and Simon (2025) show that the effects of AI attribution on trust can be affected by additional information about the sources used to produce news, while Morosoli et al. (2025) demonstrate that responses to AI labels can vary across news contexts and outcomes. These findings support the need to investigate disclosure as a specific communication practice rather than assuming that all forms of disclosure have identical consequences. For digital media users in Lagos State, it is important to establish whether awareness of AI involvement is associated with greater or lower perceived credibility and whether clear information about AI assistance and human editorial responsibility helps users make more informed assessments. Against this background, the study examines trust calibration in AI-generated news by investigating the influence of source disclosure on news credibility among digital media users in Lagos State.

1.2 Statement of the Problem

The growing incorporation of artificial intelligence into news production presents a challenge to established expectations about authorship, editorial responsibility and the credibility of published information. Traditionally, news consumers have relied on journalists, editors and identifiable media organisations to gather information, verify claims and correct errors. AI-generated news introduces a more complicated production process in which automated systems may generate text, summarise documents, organise information or contribute to other stages of content creation. Although these applications may improve speed and efficiency, they can also create uncertainty about who produced a report, how its claims were verified and who should be held responsible when errors occur. This uncertainty may affect the confidence readers place in news content, particularly when they cannot easily determine whether AI was used and how extensively it contributed to the final report. Clerwall (2014) and Tandoc et al. (2020) provide foundations for understanding the implications of automated content production for journalism and audience evaluation. The central problem is therefore not simply the use of AI in newsrooms but how audiences can make informed judgments about news credibility when the production process is partly or wholly automated.

A second problem concerns the uncertain influence of source disclosure on news credibility. Disclosure is frequently regarded as a means of improving transparency because it informs readers about AI involvement. However, greater transparency does not automatically translate into greater trust. Some readers may appreciate disclosure because it allows them to understand how a report was prepared, while others may interpret the same information as evidence that the report is less reliable or that human journalistic judgment has been reduced. Toff and Simon (2025) found that AI-generated labels could reduce perceived trustworthiness, although additional information about sources could help counteract this effect. Similarly, Morosoli et al. (2025) reported that AI attribution affected source credibility and sharing intentions in ways that depended on the news context. These findings demonstrate that the relationship between disclosure and credibility is not straightforward. The problem requiring investigation is whether source disclosure among digital media users in Lagos State improves, reduces or has no meaningful influence on perceived news credibility, and whether different forms of disclosure produce different interpretations of AI-generated reports.

A third problem is the difficulty of achieving trust calibration rather than indiscriminate trust or blanket rejection of AI-generated news. If users trust AI-generated news merely because it appears professionally written, they may overlook inaccuracies, unsupported claims or the limitations of automated content production. Conversely, if users reject all news involving AI, they may dismiss reports that have been carefully checked by qualified journalists and supported by reliable evidence. Lee and See (2004) explain that appropriate trust in automation depends on the alignment between a system’s capabilities and the user’s understanding of its reliability. Applied to news, this principle suggests that audiences need sufficient information to distinguish between different forms of AI involvement and to assess the responsibility retained by human journalists and media organisations. A disclosure statement that merely identifies AI without explaining the extent of its role or the presence of human verification may not provide enough information for such judgments. The problem is that the relationship between disclosure and credibility must be understood in terms of whether users are able to make appropriately informed evaluations, rather than whether they simply report high or low trust in AI-generated content.

Finally, there is a need for context-specific evidence concerning how digital media users in Lagos State respond to source disclosure in AI-generated news. Although international studies have examined automated journalism, AI attribution and news credibility, their findings may not fully explain how users in a Nigerian setting interpret disclosure labels. Users’ evaluations may be influenced by their prior experiences with digital news, familiarity with AI, perceptions of media organisations and the clarity of the disclosure presented to them. It would therefore be inappropriate to assume, without empirical investigation, that findings from other countries apply in exactly the same way to Lagos State. The study seeks to address this problem by examining the influence of source disclosure on perceived news credibility among digital media users in Lagos State, with attention to whether disclosure helps users understand AI involvement and form more appropriately calibrated judgments. Its central concern is to establish how disclosure relates to perceived accuracy, believability and trustworthiness, thereby contributing evidence that may assist journalists, editors and digital publishers in communicating AI use responsibly.

1.3 Aim and Objectives of the Study

The aim of this study is to examine trust calibration in artificial intelligence-generated news by investigating the influence of source disclosure on news credibility among digital media users in Lagos State.

The specific objectives are to:

1. Examine the level of awareness of AI-generated news among digital media users in Lagos State.

2. Assess the influence of source disclosure on the perceived credibility of AI-generated news among digital media users in Lagos State.

3. Examine how different forms of source disclosure, particularly AI-generated and AI-assisted disclosures, relate to perceived news credibility among digital media users in Lagos State.

4. Determine the extent to which source disclosure supports appropriate trust calibration in AI-generated news among digital media users in Lagos State.

1.4 Research Questions

The following research questions will guide the study:

1. What is the level of awareness of AI-generated news among digital media users in Lagos State?

2. How does source disclosure influence the perceived credibility of AI-generated news among digital media users in Lagos State?

3. How do different forms of source disclosure, particularly AI-generated and AI-assisted disclosures, relate to perceived news credibility among digital media users in Lagos State?

4. To what extent does source disclosure support appropriate trust calibration in AI-generated news among digital media users in Lagos State?

1.5 Research Hypothesis

The following null hypothesis will be tested at the 0.05 level of significance:

H₀: Source disclosure has no statistically significant influence on perceived news credibility among digital media users in Lagos State.

1.6 Significance of the Study

The study is expected to be significant to journalists, media organisations, digital media users, policymakers, journalism educators, AI developers and future researchers.

Journalists and editors: The findings may help journalists and editors understand how audiences interpret information about AI involvement in news production. The results could guide decisions concerning the wording, placement and clarity of disclosure statements. They may also encourage news professionals to distinguish clearly between AI-assisted content that has undergone human review and content generated largely through automated systems.

Media organisations and digital publishers: The study may provide evidence to support the development of transparent editorial policies for AI use. Media organisations could use the findings when designing disclosure labels, explaining editorial verification procedures and determining how responsibility for published information should be communicated. The findings may also help publishers understand whether disclosure practices strengthen confidence in their news products or unintentionally create uncertainty among readers.

Digital media users: The study may increase awareness of the importance of evaluating news based on evidence, source information and editorial accountability rather than relying exclusively on the appearance or perceived sophistication of a report. It may encourage users to distinguish between AI involvement and verified accuracy, thereby supporting more informed news-consumption decisions.

Policymakers and media regulators: The findings may contribute to discussions about transparency, accountability and responsible AI adoption in the Nigerian media industry. Evidence about how audiences respond to disclosure could inform the development of practical guidelines for communicating AI involvement in news production. However, any policy recommendations would need to consider the broader legal, professional and technological context.

Journalism and communication educators: The study may provide useful material for teaching media ethics, digital journalism, news evaluation and AI literacy. It could help educators explain why transparency should be accompanied by fact-checking, editorial oversight and clear accountability.

AI developers and technology providers: The findings may help developers appreciate that the design of AI-supported news systems should consider the needs of the people who consume the resulting information. Clear explanations of AI involvement, verification processes and system limitations may be important components of responsible news-related AI applications.

Future researchers: The study may contribute to the growing literature on AI-generated news, source disclosure, credibility and trust calibration. It may also provide a foundation for further studies comparing different Nigerian states, media platforms, age groups and categories of AI disclosure.

1.7 Scope of the Study

The study focuses on trust calibration in AI-generated news, specifically the influence of source disclosure on perceived news credibility among digital media users in Lagos State, Nigeria.

Content scope: The study covers awareness of AI-generated news, source disclosure, perceived news credibility and trust calibration. Source disclosure refers to information indicating whether AI generated or assisted in producing a news report and, where relevant, whether a human journalist or editor reviewed the content. Perceived news credibility concerns users’ judgments of the accuracy, believability and trustworthiness of the news presented to them.

Geographical scope: The study is limited to Lagos State. It does not claim to represent all digital media users in Nigeria.

Population scope: The study focuses on digital media users in Lagos State who encounter news through internet-based platforms. The final study population, sampling frame, sample size and eligibility criteria will be specified in Chapter Three.

Methodological scope: The study will examine users’ perceptions and responses to source disclosure. Depending on the adopted research design, the investigation may use a structured questionnaire, controlled exposure to news items with different disclosure labels, or a combination of appropriate methods. The methodology should be selected to ensure that the research questions and hypothesis can be addressed adequately.

1.8 Limitations of the Study

The study may encounter certain limitations that should be considered when interpreting its findings.

First, self-reported responses may not always reflect actual news-consumption behaviour. Participants may indicate that they trust or distrust AI-generated news differently from how they behave when encountering such news in everyday digital environments.

Second, differences in AI literacy may influence how participants understand disclosure labels. Some users may already be familiar with generative AI, while others may have limited knowledge of how AI contributes to content production. These differences may affect their credibility judgments.

Third, variation in prior media trust may influence responses. Participants who already distrust news organisations may react differently to disclosure than those who have confidence in established publishers. Where feasible, the study should measure relevant background characteristics and consider them when interpreting the results.

Fourth, the wording and placement of disclosure statements may affect participants’ responses. A label that explains AI assistance and human verification may be interpreted differently from a general statement that a report was generated by AI. The study should therefore define its disclosure conditions clearly.

Finally, the geographical focus on Lagos State may limit the generalisation of the findings to other parts of Nigeria. Differences in digital access, media habits, educational experiences and familiarity with AI may influence responses elsewhere. These limitations will be acknowledged when presenting and discussing the findings.

1.9 Operational Definition of Terms

Artificial Intelligence (AI): Computer-based technologies capable of performing tasks commonly associated with human intelligence, including language generation, information processing, summarisation and content analysis.

AI-generated news: News content produced substantially or entirely by an AI system, with or without human editing or verification.

AI-assisted news: News content in which AI supports some aspects of production, such as research, summarisation, drafting or editing, while a human journalist or editor retains a role in preparing or approving the final report.

Source disclosure: Information presented to readers to identify AI involvement in producing a news item. In this study, it includes disclosure that content was generated by AI or prepared with AI assistance, as well as relevant information about human editorial review where applicable.

News credibility: The extent to which digital media users perceive a news report as accurate, believable and trustworthy.

Trust calibration: The process of developing a level of confidence in AI-generated news that is reasonably aligned with the evidence available, the system’s limitations, the degree of human verification and the accountability of the publisher.

Digital media users: Individuals in Lagos State who access or consume news through internet-enabled platforms, including news websites, online newspapers, social media and other digital channels.

Source credibility: The perceived trustworthiness and competence of the person, system or organisation presented as responsible for producing or publishing news.

Human editorial oversight: The involvement of a human journalist, editor or other responsible professional in reviewing, verifying, correcting or approving AI-supported news content.

1.10 Organisation of the Study

The study will be organised into five chapters.

Chapter One presents the introduction, background to the study, statement of the problem, aim and objectives, research questions, research hypothesis, significance, scope, limitations and operational definitions of terms.

Chapter Two reviews relevant conceptual, theoretical and empirical literature on artificial intelligence in journalism, AI-generated news, source disclosure, news credibility, trust calibration and digital media consumption. It will also identify the research gap and present the theoretical framework guiding the study.

Chapter Three explains the research methodology, including the research design, population, sample and sampling techniques, research instrument, validity and reliability procedures, methods of data collection and techniques of data analysis.

Chapter Four presents, analyses and discusses the data collected in relation to the research objectives, research questions and hypothesis.

Chapter Five presents the summary of the findings, conclusion, recommendations, limitations of the study and suggestions for further research.

Project – Trust Calibration in Artificial Intelligence-Generated News: The Influence of Source Disclosure on News Credibility Among Digital Media Users in Lagos State.