## I. INTRODUCTION Artificial intelligence (AI) is transforming the media industry on a global level. Whether it's automatic news writing or custom recommendations, AI is now changing the way we produce, distribute, and consume information. Today, media organizations are able to produce stories rapidly and economically (both business, sports, and elections) thanks to natural language generation (NLG) vehicles and machine learning algorithms (Carlson, 2020). Meanwhile, recommendation engines used by services like Facebook rely on behaviors which makes information feeds on these services' dependent on the behavior of users, which leads to higher productivity and engagement of the users but also makes this technology raise important ethical questions (Pariser, 2011; Graefe, 2016; O'Neil, 2016). A key issue is that algorithmic decision-making is not transparent. Popularized as "black boxes," AI programming systems operate on large quantities of data and offer little explanation as to how or why certain material is selected (O'Neil, 2016). Accountability is thus denied without transparency and it becomes hard for the user to critically assess the information sources credibility. When trained with biased data, algorithms have the power to propagate stereotypes and political divisions (Binns, 2018). However, as increasing online use of engagement-oriented metrics determines the flow of content, accuracy is trade off with sensationalism and a further acceleration in the diffusion of misinformation and disinformation (Bakir & McStay, 2018; Vosoughi et al., 2018). The question of ethics is even more grave in Bangladesh, where fragile institutions, deep cleavages, and low digital literacy make people more vulnerable to misinformation. More than 100 million internet users were identified by 2023 and social media became the biggest news source for this large group of population (Islam & Rahman, 2019). However, the design of Facebook's algorithm tends to give a higher visibility to emotionally charged and misleading content as well, contributing to political manipulation, sectarian narratives and health-related falsehoods. And in some cases, there have been dynamics that have not only undermined the social cohesion but also forced the outbreak of violence (Islam & Rahman, 2019). On 20th April 2024 a fake video about Bangladesh's Prime Minister apologizing for what had happened and swearing under a caretaker government circulated over Facebook. Created by artificial intelligence (AI), the video went viral before fact-checkers were able to interfere. This was not simply a rumor for many; it changed political opinion in an extremely divisive election year. While Al-based misinformation is a widely researched issue in the Global North, the research in Bangladesh and the Global South which has taken place so far has mainly been descriptive in nature mapping the false news patterns or highlighting general regulatory problems. What has been missing is a systematic empirical examination of how AI has been integrated into journalistic work processes and content streams and how these automation processes raise specific ethical issues. Unlike prior reports, this study is an empirical case studies of AI-created or altered content, as well as deep interviews of Bangladeshi media actors in order to provide rare insights in intersection of technology, ethics and journalism in otherwise a fragile democracy. Although YouTube, TikTok, WhatsApp and other social media platforms are gaining popularity in Bangladesh, Facebook is still the absolute source for information and news with more than 30 million active monthly users. It is the primary source for citizens to consume mainstream news and generated content. It has been cited as the source for misinformation during elections, health crises and religious conflicts, as the Facebook algorithm is intended to maximize engagement rather than accuracy. Because of this, Facebook is the most pertinent article to be studied with regards to ethics of AI in relation to journalism and public communication in Bangladesh. The present study critically analyzes delete the word discourse study critically discourses the ethical issues brought up by the AI-driven media practices (managing information in news sharing) in consideration of its primary platform Facebook, the world's largest social networking site. By using interviews with fact checkers, journalists, technologists and editors as well as case studies of AI-generated misinformation, the study reveals how AI technologies both assist and restrict democratic communication. ## II. RESEARCH PROBLEM Artificial Intelligence (AI) has also introduced mighty tools of content creation, curation, and dissemination, but with severe ethical issues relating to bias, disinformation, transparency, and accountability (Graefe, 2016; O'Neil, 2016; Vosoughi et al., 2018). While these issues have been discussed of intense study in the Global North i.e., during the 2016 U.S. elections, Brexit, and GDPR debates research interest in the Global South is thin (Bakir & McStay, 2018; UNESCO, 2020). This opens up questions about the impact of AI on journalism and public communication in weakly regulated and polarized democracies. Bangladesh can be considered as a good example. Facebook dominates the country's information landscape with millions of people receiving their information mainly from it. Its engagement-maximized algorithm has propagated sensational and partisan material, especially during and after the 2024 national elections, when deepfakes and fabricated news were common. While global reports indicate simultaneous threats, few empirical studies have addressed the way AI technologies are embedded within Bangladeshi news work, the way ethical dilemmas emerge in real-world newsrooms, or the way institutions seek to respond. This work is mostly descriptive, and theoretical frameworks like Media Ethics and Technological Determinism are still under-studied here. In addition, legislative measures such as the Cyber Security Act (2023) fail to address algorithmic bias, deepfakes, or platform accountability with satisfactory measures. Herein, this study examines: (1) the ways in which AI is being integrated into news practices in Bangladesh, especially on Facebook; (2) the most significant ethical issues it presents; (3) institutional, regulatory, and fact-checking reactions; and (4) potential frameworks for ethical, legal, and technological reforms. ## III. SIGNIFICANCE OF THE STUDY This paper offers a critical insight to the existing discourse on artificial intelligence and media ethics by presenting empirical intelligence drawn from places. In this case, Bangladesh, a democratic country within the Global South where weak institutions and a lack of political consensus heightens algorithmic risks. This work not only leaves the surface observation of how media practices are operating but by conducting systematic interviews with working journalists and by examining specific examples of artificial intelligence transforming journalism and the overall course of public discourse, the study successfully demonstrates the substantive nature of the problem of artificial intelligence beginning to transform journalism and society-wide discourse as a whole. The Media Ethics Theory and Technological Determinism provide theoretical foundation, which together make possible a considerably fine-grained deconstruct of the obviously evoked ethical breaches and the more enduring structural consequences spawned by Al-assisted systems. Empirically, the investigation makes new knowledge about dynamic interaction between algorithmic infrastructures and institutional latency, as well as lacks digital literacy. In practice, the research outlines the tangible plans of transformational change, such as coming up with ethical codes, internalizing legal protection measures, and licensing tech-based solutions aimed at aligning AI towards positive communication in democratic states. ## IV. LITERATURE REVIEW The utilization of AI in the production, editing, and shipping of media has triggered radical changes in the creation and consumption of information on the world level. By virtue of its ability to automate processes, to tailor content to individual users and fine-tune delivery mechanisms, artificial intelligence has altered without doubt the dynamics between media-entities and their audiences. However, for as clear and advantageous these outcomes are, they come with serious ethical considerations namely those of bias, misinformation and accountability issues (Graefe, 2016; O'Neil, 2016). Within the Bangladeshi context, these determinants take on an especially strong relevance, when considering the fact that the role of AI-enabled platforms in organizing both the political debate and the social exchange has gained a particular salience. ### a) Global AI-Media Dynamics In the 21st century, artificial intelligence penetrated the field of journalism, and appeared at first in applications for automated sports journalism and data-driven journalism (Graefe, 2016). Over the past few years, the technological space has grown to include natural language generation (NLG), algorithmic curation and automation of video production, and personalized information ecology - as exemplified by Facebook and YouTube - have completely changed the information ecology by using behavioral data to curate the information people see (Pariser, 2011). The global market of AI-powered content creation and recommendation reached over USD$1 billion in 2018 and this marked a large commercial success (Binns, 2018). In spite of these tremendous advancements, there have constantly been challenges that still persist to date. AI models built on biased datasets are prone to reinforce and even magnify harmful stereotypes and thus contribute to increasing social inequality and political polarization (Noble, 2018; Diakopoulos, 2019). Models for engagement maximization almost always favor emotionally charged or sensational content, and hence are more likely to accelerate information spread, especially in the case of misinformation (Vosoughi, Roy, & Aral, 2018). With recent media attention, the 2016 US presidential election and the Brexit, it has been highlighted that algorithmic content curation can erode democratic institutions by spreading false/inflated information on a large scale (Bakir & McStay, 2018). These issues are further complicated by the black box nature of participating AI systems, which fails to reveal the methods of ranking used in their operation, and thus reduces transparency and accountability (O'Neil, 2016). ### b) Global Responses to Ethical Challenges Global attempts to avert such ethical problems have been uneven. Algorithmic transparency is one of the explicit requirements of the European Union's General Data Protection Regulation (GDPR), and it states individuals' right to challenge the results of automated decision making (Floridi et al., 2018). The recent push by academics and others has made the case for stronger ethical guidelines that stress fairness, openness and inclusiveness (Binns, 2018; Diakopoulos, 2019). Organizations, such as UNESCO, have also driven the need to hold algorithms accountable for safeguarding democracy (UNESCO, 2020). Nevertheless, dominant tech platforms (Facebook, YouTube and Twitter) continue to use engagement-focused algorithms that contribute to the problem of polarization and misinformation on a global scale (Tufekci, 2017; Benkler, Faris, & Roberts, 2018). ### c) AI in Bangladesh's Media Ecosystem In Bangladesh, the adoption of artificial intelligence in journalism is in the starting phase. The biggest influence currently comes to see in social media such as Facebook originates as the main source of news for millions of users. Its AI-driven algorithms curate content feeds and as such, alter how and what users' access and interpret news (Islam & Rahman, 2019). Although some news outlets have focused on artificial intelligence in order to improve financial reportage or electoral analysis, the use of large-scale automated systems is still relatively small (Rahman, 2021). The polarized politics and partisan media environment of the country increases the threats of algorithmic curation. AI systems often perpetuate echo chambers, whichever the case, aggravate political divisions and speed up the spread of disinformation (Hossain Haque, 2020). During the national elections, Al-propaganda and manipulated content have shaped the perceptions of voters and incited unrest (Islam & Rahman, 2019). ### d) Ethical and Institutional Challenges in Bangladesh Although Bangladesh has significant potential, the country faces tremendous challenges in how it can best use the food aid. Local vulnerabilities are multiplying the global problems of Al. Biased Training DataSets: Political and Social Prejudices Are Embedded in Al-Generated Content Hasan, Emanuel (2020) Engagement-fuelled algorithms perpetuate the existing power structures by subscribing to the politically biased material (Benkler, Faris, & Roberts, 2018). Low digital literacy adds to the problem, and leaves citizens incapable of separating credible information from misinformation. Notable examples of this are the misinformation of health in the context of the Covid 19 pandemic, and politically motivated deepfakes during elections (Islam & Rahman, 2019). Institutional responses are still not good. The Digital Security Act of 2018 takes cybercrime and misinformation into account but fails to adequately address algorithmic bias and does not offer content curation transparency (Hasan, 2020). Fact - checking organizations, such as Bangladesh Media Council, try to debunk false claims, but lack the technological capacity to keep pace with AI - driven disinformation (Rahman, 2021). This gap between the lawful structure and high-tech realities implies the need for stronger and fuller responses. ### e) Towards Ethical, Legal, and Technological Frameworks There is a strong agreement in literature on the need to design a comprehensive response towards curbing the negative impacts of artificial intelligence within the media space. Ethically, transparency, equity, and inclusivity should be the central principles of the AI systems. The lawmakers must make the digital platforms reveal their content-modernization policies, the user have a means of appealing their algorithmic decisions, and corporations be responsible in the future when wrong information spreads (Floridi et al., 2018). The technological solutions like Explainable AI (XAI) and AI-based fact-checking systems are also promising as far as they can increase the level of transparency and prevent the spread of fake pieces of information (McStay, 2018). The media environment is especially susceptible in such a situation like the case of Bangladesh when the institutional capacity is low, which is why such protective measures are essential. Without them, AI will threaten to divide people along political lines, destroy the trust that people have in journal intervention, and even subvert the very nature of a democratic discourse. In this regard, an all-encompassing remedy through ethical advice, legislative accountability, and the latest advancements in technology is not only necessary, but essential, so that AI application reinforces and never compromises the interest of the general population. ## V. THEORETICAL FRAMEWORK The Bangladeshi media culture reveals an essential context in which one can examine the moral dilemma of AI in media manufacturing and sharing. In this paper, two main theoretical perspectives Media Ethics Theory and Technological Determinism were used to unravel the intricate layers involved in the relationship between journalism and artificial intelligence. Such frameworks have been chosen on the basis of their ability to cover both the ethical justified imperatives to which journalistic practice is underpinned and the widespread structural implications of technological intervention. Although other models would have provided some peripheral relevance like Agenda-Setting, or the Democratic-Participant Theory, the two emphasize the dynamics of participation, as well as, the opinion of the people in general, not focusing on the two forces of ethicality and technology involved. On the contrary, both Media Ethics Theory and Technological Determinism can help look deeper into not only the ethical failure but also the determinism behind the misinformation provided by AI. # a) Media Ethics Theory The Theory of Media Ethics offers a standard by which it is possible to judge the ethical aspects of Al-based curated journalism. The most important principles of this framework are truth, responsibility, and justice (Christians et al., 2019). The theory assumes that the greatest task of journalism is to fulfil the interest of the people by means of honesty, fairness, and accountability. In an informational sector that is already full of misinformation and lies, like Bangladesh where fake news resorts to epidemic rates, this theoretical lens can provide a strict pathfinder on how AI tools can enhance journalism or diminish it. An example of such a dynamic is Facebook: the labels and hashtags can be used to reach more people, but its algorithms provide more and more priority to those pieces of content that create the greatest engagement, and, in most cases, it favors sensational or biased coverage (Bakir and McStay, 2018). This prioritization brings down to the level of the media its responsibility to inform the citizenry in a reliable and trustworthy manner. In addition, an opaque nature or black box (O'Neil, 2016) of algorithmic architecture makes accountability mechanisms even more complicated. Using the Media Ethics Theory, this article outlines the consequences linked to the rising usage of artificial intelligence during news generation as well as clarifying how it can corrupt the ethical standards that define responsible news coverage. ### b) Technological Determinism: AI as a Driver of Social Change Technological determinism gives a structural approach, and claims that technology is never a neutral background but rather, an active process and influences the creation of social norms, behaviours and a political reality (Smith and Marx, 1994). In the world of Bangladesh, where there is so much misinformation and political prejudice, Al-based algorithms on social media have a significant power on the information space, thus defining and organizing the systems of beliefs. These algorithms are systematic to categories of content, which systematize the content according to the previous interactions of the user, thus reinforcing the existing biases and increasing political differences (Benkler, Faris, and Roberts, 2018). Measures like clicks, shares and dwell time facilitate this phenomenon at the cost of veridical information towards viral information. Although the user might see agency in the decision-making of reading, the so-called algorithmic filter bubbles (Pariser, 2011\) will cunningly lead the user in the direction of other opinions. Technological determinism, therefore, not only explains the shaping of the popular speech into AI systems but also their opposing part, creating echo chambers and preventing the exposure to a wide range of opinions. ### c) Integrating Media Ethics and Technological Determinism As a combination, these two theoretical lenses create a broader analytical model of studying the effects of AI on the media and its ecosystem in Bangladesh. The Theory of Media Ethics anticipates the gap between the workings of the algorithms and the principles of journalism, including truth, fairness, and accountability. Concurrently, Technological Determinism is used to outline the channels, within which these technologies influence the discourse of the population and divide political groups. With its integrated framework, a complex approach can be made to analyzing both micro-level ethical issues, truth, accountability, and macro-level structural issues, polarisation, algorithmic hegemony. The two theoretical stances have different solutions that allude to: The Media Ethics Theory holds strict and codified ethical norms and strict regulation, and Technological Determinism proposes wholesale systemic reform, including effective regulation infrastructure and a radical re-coding of the digital platform. Combined, these observations confirm that pressing back against AI mediated disinformation in Bangladesh requires two commitments that lie apart but complement each other, namely a commitment to moral rectitude, and a commitment to bring about structural change. ## VI. METHODOLOGY This study has got a qualitative multi-method design, which is a combination of semi-structured interviews and in-depth case studies. The central goal is to explore the implementation of artificial intelligence in Bangladeshi journalism and challenge the ethical issues, such as prejudice, misinformation, lack of transparency, and lack of accountability, in this scenario. The research timeframe is 2020–2025 which coincided with national elections and the July Uprising in 2024 when AI generated content played a crucial role in fueling an already polarized information space. ### a) Sampling Strategy Twenty respondents with first-hand experience in the development or operation of media AI content were sampled using purposive, criterion-based sampling. The sample was balanced, including five editors, five journalists and five media producers, five communication or AI specialists, which led to a wide variety of different perspectives and yet at the same time a small enough sample size to handle in-depth thematic analysis. Snowballing was then employed to get other respondents who could fit the eligibility requirements, including newsroom heads or AI consultants. For the case studies, a collection of twenty examples of Al-generated or assisted content have been collected from distinguished online news channels and social networks from Bangladesh. Selected cases were focused on high impact areas, like politics, society, and health. Five cases were finally selected using a combination of the critical case logic (emphasizing most influential) and maximum variation logic (reflecting diversity). The resulting dataset allowed the identification of common trajectories and at the same time enabled the rich contextual information of each selected case. ### b) Data Collection Data collection was done on two parallel streams. - Semi-structured interviews were conducted to investigate the use of AI in journalism particularly, automated content generation, algorithmic curation, moderation and machine translation. They also covered perceived value and drawbacks, ethical concerns over bias and misinformation, organizational instructions, and the sentiments over regulation and people's trust. Interviews were carried out in face-to-face and via secure video conferencing. Participants got informed consent and were assured of the confidentiality of their responses and of the right to withdraw. Each interview was between 45-60 minutes in length; recorded with consent, transcribed precise, and decoded into English when needed. - The case study was conducted based on news articles, social media posts, videos, images (including fake media), and explanatory content in which AI involvement was clearly stated or from which AI involvement fairly could have been reasonably inferred. Inclusion criteria Our data collection was limited to factually verifiable evidence of AI authorship or enhancement, such as newsroom activities, platform labels, correction notices, or fact-checks by third parties. For each item, all relevant case-specific metadata, including, time of publication, interaction, correction, and popular responses, was collected systematically. ### c) Data Analysis The interview transcripts were conducted using reflexive thematic analysis. Coding was done in iterative loops of open, focused and axial code generation. The interview guide and in-vivo terms were used as seeds and a preliminary codebook was developed from dual coding of a representative sample of transcripts. Discrepancies were discussed to develop the intercoder reliability. A comparative matrix and analytical memos were used to perform cross-role analysis between editors, journalists, technologists and academics. Qualitative content analysis was used to analyse the collected data. Cases were reviewed independently and then combined using cross case synthesis. The analysis revealed four different thematic categories: (1) typologies of AI intervention (2) claimed ethical problems; (3) situational contexts of intervention proposals, elections, crises, and health emergencies; and (4) reactions and responses from organisations and platforms. Subsequently, the exegesis and the pattern-matching procedures were used to synthesize the findings of the sample. ### d) Validity, Ethics, and Reflexivity The research used methodological triangulation (interviews and case study research), data source triangulation (professional positions, media outlets), and data-type triangulation (text, image, video, metadata) in order to be more valid. This multi-level approach added strength and credibility to the findings. Ethical behavior was one of the principles of the research. All informed consent was obtained from all interview respondents and where applicable, identifying information was anonymized. Data were stored in secure storage databases. The research adhered to ethical principles for qualitative research and permission was secured from institutions before the data collection. Reflexivity was integrated in the research process. The author had a reflexive journal that recorded assumptions, positionality, and decision-making processes to create awareness of possible biases when interpreting politically sensitive content. Peer debriefing with other academic colleagues was used as an additional tool to increase transparency and critical reflection. ### e) Methodological Limitations As it does with most qualitative investigations, this study has limitations. First, while a group of twenty participants were able to reach the point of thematic saturation, this may be insufficient to reflect the diversity of experiences across the long Bangladesh media landscape. Second, the use of self-reported interview data makes the results prone to a number of biases, including selective recall or socially desirable responding; the addition of case studies partially addresses this issue. Third, the selection of case studies was limited, by necessity, to those incidents that were documented or traceable and this may have led to the omission of less visible or undocumented instances of Al utilisation. Finally, given that the problem of Al driven misinformation is a politically sensitive subject in Bangladesh, ethical limits may have been a cause of openness of some participants. Despite all these limitations, the inclusion of several research methods, triangulation of data sources and a commitment to reflexivity significantly add to the trustworthiness and validity of the study. ## VII. RESULTS AND DISCUSSION The results are organized according to the research questions of the study and are interpreted in accordance with the Media Ethics Theory and the Technological determinism. The analysis highlights the relationship between AI-powered processes (especially deployed on Facebook) and misinformation, bias and non-accountability in the media domain within Bangladesh. Given the illustrations based on political events, religious controversies, health communication, and crisis situations, the significant impact of AI technologies in the public discourse was found in a sensitive period for political communication. a) Research Question 1: How is Al currently being used within Bangladeshi news production and distribution, particularly on Facebook? ## i. AI-Driven Content Curation on Facebook Facebook has a roster of active monthly subscribers of over thirty million in Bangladesh, and it keeps outstripping other platforms of digital media, as it still holds a stronger status in the informational space. The algorithm which feeds the ranking system at the core of the platform favors user engagement (likes, shares, comments) over factual accuracy and ends up promoting sensationalist content that proves popular in terms of emotion, offering a fertile ground for misinformation when part of political unrest, health danger, and religious conflict. Reféred Media Ethics Theory (Christians et al., 2019), it can be seen that such algorithmic practices destroy the foundations of truth and accountability that are the cornerstones of responsible journalism. ## ii. Case Studies of AI-Generated Misinformation Data from Rumor Scanner Bangladesh Implies how Al was repeatedly used in 2024-2025 to spread politically charged and socially unruly disinformation. The examples found are deep-faked videos of political leaders, fake resignation letters, memes with doctored news images and fabricated written articles. The most important conclusions are summarized below. Table 1: Documented Cases of AI-Generated Misinformation in Bangladesh (2024–2025) <table><tr><td colspan="2">Category</td><td>Examples (Date)</td><td>AI Involvement</td><td>Impact on Public</td><td>Source</td></tr><tr><td colspan="2">Political Misinformation</td><td>Travel embargos on Sheikh Hasina &amp; Army Chief (Mar 2024); Counterfeit issue of Mirza Fakhrul (Mar 2024); Hasina apology &amp; caretaker govt. (Apr 2024); GM Quader resignation (Jun 2024)</td><td>Al-generated posts, fake news</td><td>Public confusion, political instability, distrust in leadership</td><td>Rumor Scanner</td></tr><tr><td colspan="2">Manipulated Visuals</td><td>Reformed Jugantor photo: Jamaat/Shibir arrests (Jun 2024); Edited video: Sarjis Alam misconduct (Jul 2024); Prothom Alo photocard of Dr. Shafiqur Rahman (Sep 2024)</td><td>Al-edited photos &amp; videos</td><td>Reputational damage, heightened conflict</td><td>Rumor Scanner</td></tr><tr><td colspan="2">Targeting Political Leaders</td><td>Jatiya Party leaders attacking Quader &amp; Huq (May 2024); False Dhaka College Chatrodol convener post (Jul 2024); Fabricated resignation of Advisor Asif Nahid (Sep 2024)</td><td>AI-manipulated posts &amp; content</td><td>Polarization, unrest in political/academic circles</td><td>Rumor Scanner</td></tr><tr><td colspan="2">Fabricated Articles</td><td>False news article by "Fuad" with misleading headlines (Aug 2024); False speech claim on Sarjis Alam (Oct 2024)</td><td>AI-generated text content</td><td>Misinformation in public discourse</td><td>Rumor Scanner</td></tr></table> These cases validate how AI in Bangladesh is not just general but a core part of news spreading, with algorithms boosting fabricated stories, manipulated visuals, and false narratives. ## iii. Theoretical Integration In such a way, the empirical patterns prove the fact that AI is a dynamic discursive agent, but not a passive tool. Media Ethics framework also helps to explain how the truthfulness/responsibility principle is violated when the false resignations or apologetic words attain too much media attention. The trendy arrangement (Smith and Marx, 1994) explains the effect of algorithmic engagement in controlling discourse to create an echo chamber and reinforce the rift existing in societies (Pariser, 2011). These perspectives validate how AI-enabled fabrication in Bangladesh is a systemic result of algorithmic logic interacting with flimsy democratic institutions. b) Research Question 2: What critical ethical issues do Al-driven media practices raise, bias, misinformation, and opacity, accountability, and fairness? ## i. Amplification of Misinformation and Bias Since they are treated like mere cultural attributes, the security team can readily amplify misinformation and bias. There are consistent cases of ethical breaches in Table 1 of RQ1. The fake news and fake images as well as stories told about politics were given priority by Al-driven algorithms, thus taking engagement metrics over factuality. In terms of the Media Ethics Theory (Christians et al., 2019), this is an apparent violation of journalism and its duty to tell the truth, be fair, and be accountable. ## ii. Case Studies of Ethical Failures These cases are not to be treated as the chronicles of solitary cases but rather they can be viewed in the prism of moral integrity that they are violating. Table 2: Ethical Issues in AI-Generated Misinformation in Bangladesh (2020–2025) <table><tr><td>Category</td><td>Examples (Date)</td><td>AI Involvement</td><td>Impact on Public</td><td>Source</td></tr><tr><td>Political</td><td>Opposition contenders dropping out (2024); Hasina apology &amp; caretaker govt. (2024); GM Quader resignation (2024); Leaders beating Quader &amp; Huq (2024)</td><td>AI-generated posts, fake content, manipulated visuals</td><td>Voter confusion, political polarization, undermining of democratic processes</td><td>Rumor Scanner</td></tr><tr><td>Health</td><td>Herbal medications as COVID-19 cures (2020)</td><td>AI-generated videos</td><td>Public health risks, spread of dangerous practices</td><td>WHO (2021); Friedrich Naumann Foundation</td></tr><tr><td>Religious</td><td>Forged images of Quran desecration (2021)</td><td>AI-manipulated photos &amp; videos</td><td>Religious tensions, incitement to violence</td><td>VOA (2024); Fact Check Bangladesh</td></tr><tr><td>Disaster</td><td>False flood reports during Cyclone Mocha (2023)</td><td>AI-enhanced videos &amp; photos</td><td>Public panic, unnecessary evacuations</td><td>DismissLab (2023); Dhaka Tribune</td></tr></table> These examples show how algorithmic bias extends political gaps, how misrepresentation spreads during crises, and how accountability gaps put citizens at risk. ## iii. Theoretical Integration From the standpoint of the Media Ethics Theory, these incidents represent systematic violations of the principles of the truth, justice and responsibility. Concurrently, Technological Determinism allows for some explanation of their recurrence: the filter bubbles and the over-enhancement of divisive narratives are Facebook's algorithmic architecture (Pariser, 2011). Cases where fabricated resignations, fake cures for the coronavirus and images of Quran desecration have been discovered show that these harms are not unintentional, but embedded in the way the algorithmic prioritization logic functions. Overall, the findings suggest that the deliberation of AI-driven misinformation in Bangladesh is an ethical lapse and the outcome of the design of technology. Bias fortification undermines diversity, misinformation damages trust, and opaque algorithms hinder liability. c) Research Question 3: How are regulators, fact-checking units, and media organizations in Bangladesh addressing AI-generated misinformation? ## i. Institutional Responses and their Limits Bangladeshi organizations recognize the dangers of Al-fueled news, but responses are scattered and mostly mercurial. Fact-checking organizations like Factcheck Bangladesh regularly debunk manipulated images, deep-fakes, and fake news, but their response to the issue usually comes much later once the spread of false content has already taken place. Media enterprises monitor these misinformation trends and launch misinformation awareness campaigns, which, due to limited resources and a lack of tools for detecting Al misinformation, are limited. Regulatory bodies, particularly the Bangladesh Telecommunication Regulatory Commission (BTRC), have issued public warnings, but enforcement is weak without an Al-specific legal framework. The Digital Security Act (2018) has been used to fight disinformation when the country has been in unrest, but there is no mention of algorithmic transparency or risk of misuse. Table 3: Institutional Responses to AI-Generated Misinformation in Bangladesh <table><tr><td>Actor</td><td>Response Strategy</td><td>Limitations/Gaps</td><td>Illustrative Examples</td></tr><tr><td>Fact-checking groups</td><td>Debunking false claims; publishing corrections on websites and social media</td><td>Reactive rather than proactive; corrections often slower than viral spread</td><td>Rumor Scanner corrections during 2024 election misinformation</td></tr><tr><td>Media outlets</td><td>Reporting on misinformation trends; conducting awareness campaigns Limited resources</td><td>Limited resources and training; lack of advanced AI detection tools</td><td>News coverage of deepfake incidents in political crises</td></tr><tr><td>Regulators (BTRC)</td><td>Providing public warnings about deepfakes and disinformation; monitoring online activity</td><td>No AI-specific legal framework; lack of technical expertise; weak enforcement capacity</td><td>BTRC advisories during 2024 election disturbances</td></tr><tr><td>Government</td><td>Invoking of Cyber Security Act 2023 (CSA) against misinformation</td><td>Broad scope risks misuse; Act does not directly address AI-related challenges</td><td>Use of CSA during political unrest, but no AI-specific provisions</td></tr></table> Table 3 shows, institutions are active, but their approaches are structurally one-sided with the speed and scale of algorithmic distortion. Therefore, from a Media Ethics Theory perspective, there is a debasement of truth and responsibility, and from a perspective of Technological Determinism, the emerged delay relates how the Opaque, consumption-driven algorithms dictate terms of visibility and vitality outside institutional norms. ## ii. Theoretical Integration The responsibility gap is both ethical and organizational. Institutions attempt to embody the responsibility of truth, but are unable to keep pace with the velocity of the algorithmic systems. Apart from technological determinism, the reason for what appears as weak regulation is mostly the result of deterministic design. Precisely because AI-driven platforms contain architecture that is much more agile than the tools and frameworks that citizens use, the citizens remain vulnerable. d) Research Question 4: What ethical, legal, and technological measures could be developed to regulate AI-driven media while safeguarding journalistic integrity? ## i. Towards Ethical, Legal, and Technological Frameworks Interviews with media professionals, fact-checkers, and regulators revealed a broad consensus: piecemeal responses are inadequate, and an integrated framework is necessary. Proposed measures cover ethical, legal, and technological areas. Table 4: Proposed Frameworks for Addressing AI-Driven Misinformation in Bangladesh <table><tr><td>Framework</td><td>Key Proposals</td><td>Expected Outcomes</td></tr><tr><td>Ethical</td><td>Label AI-generated content; run public awareness campaigns; enforce newsroom standards for responsible AI use</td><td>Greater transparency; improved public trust; reduced spread of false content</td></tr><tr><td>Legal</td><td>Enact AI-specific legislation; mandate disclosure of algorithms; penalize platforms for harmful content; regulate deepfakes</td><td>Platform accountability; control of disinformation flows; protection of democratic processes</td></tr><tr><td>Technological</td><td>Develop real-time AI detection tools; adopt Explainable AI (XAI); integrate AI-based fact-checking into platforms</td><td>Early identification of false content; algorithmic transparency; stronger user confidence in media</td></tr></table> As it is shown in Table 4, the introduction of stricter ethical principles significantly improves transparency in journalism; introducing specific legal changes contributes to better accountability in the digital environment growth; and the introduction of new sophisticated technological devices provides proactive protection opportunities. ## ii. Theoretical Integration In the view of the Media Ethics Theory, the proposals that were subjected to the examination confirm the principles of the truth, fairness, and responsibility since they required labeling disclosure, labeling, and citizen awareness. Technological Determinism, however, explains the need behind the structural interventions; i.e., real-time detection algorithms and explainable artificial intelligence; since the algorithms themselves influence the discourse of the people by creating a filter bubble and echo chamarone (Pariser, 2011). Taken together, these steps establish a logical roadmap to putting the artificial intelligence back in line with democratic values; the principles of ethics that form the navigating star, the legal changes that ground the responsibilities, and the technological development that provides the tools that allow intervention with the algorithm. ### e) Integrative Discussion When summarized the whole result, the results obtained in all the four research questions give a clear indication that artificial intelligence has become a powerful tool in shaping the media landscape in Bangladesh. Not only does it enhance the distortion of facts but it also threatens to harm the ethical, institutional, and democratic principles of communication. The initial research question shows that Al-filtered content on Facebook has a tendency of prioritizing sensational stories over factual coverage and thus disarming the main purpose of journalism. The second question illustrates that misinformation does not just exist in the political field but also in the knowledge of the populace about health, religion, and disaster events and reveals a severe lack of truth, responsibility, and justice. The third question shows the failures of institutional reactions: factcontextualizers, media corporations, and regulation organizations are frequently too slow and they lag behind the fast-changing technology that leaves a significant responsibility intersection. The fourth question reminds of the pressing necessity of comprehensive solutions, including ethical, legal, and technical ones, that will both help overcome the ethical and structural problems of AI in media. Using the Media Ethics Theory and Technological Determinism, this paper demonstrates that the AI in Bangladesh is anything but strictly neutral. Instead, it aggressively produces discourse, which requires rational ethical changes and whole system redresses. This synthesis forms the basis of the final statements, in which the study has contributed to the theory, to empirical knowledge and to real-life reform synthesized. ## VIII. CONCLUSION This study explored the implications of artificial intelligence created and curated content on journalism and overall communication to the population in Bangladesh, especially the most popular source of information, which is Facebook. Using a combination of interviews with editors, journalists, technologists and fact-checkers and case studies about AI-created or manipulated content, the current study reveals that engagement, an algorithmic performance measure is achieved at the cost of accuracy, which increases bias, misinformation and polarization. The findings indicate that the clouded systems of Facebook are broadcasting sensationalist and partisan materials, compromising the ethics of truth, responsibility and fairness. The case study provides empirical findings that reveal the tendency of misinformation that is repeated with political, health, religious, and disaster coverage and especially devastating consequences that were witnessed in the 2024 national elections. Reactions of regulatory authorities, fact-checking agencies and the press, have, nevertheless, been disjointed and reactive, with little to no actual responsibility. In theory, the paper utilizes Media Ethics Theory to examine the degree to which journalistic principles are violated, and Technological Determinism to demonstrate the overall effect of AI on the discussion among the audience. It is empirically enlightening about the interaction between the algorithmic systems, weak institutional organizations and the politically polarized society. In a more practical sense, the suggested framework is a cozy continuum of solutions based on ethical standards, legal changes and technological instruments, which are all designed to fight the increasing challenge of AI-induced misinformation. Overall, the analysis shows clearly that artificial intelligence is not an objective technology but it constructs information flow in the fragile democratic establishments. Being the first empirical systematic investigation of AI-enhanced misinformation in Bangladesh, this work presents evidence and practitioner recommendations, which could be used to develop the architecture of AI systems and support democratic values in reestablishing population confidence. ### a) Practical Implications The study outlines major strategies for addressing the risks of misinformation mediated by AI in the fragile media setup in Bangladesh. #### For Media Organizations - Institute and perform stringent policies that govern how artificial intelligence can be used in the newsroom. - Content created by artificial intelligence is clearly distinguishable. - In upholding editorial standards, focus on fact accuracy should be given priority to the goal of engagement metrics. - Facilitate continued training in artificial intelligence ethics and detection of mis-information. #### For Regulators - Enact AI specific legislation through Bangladesh Telecommunication Regulatory Commission or BTRC. - Platforms must be entirely transparent about the algorithms used in content recommendation. - Companies must be fully accountable for the machine learning systems that enable the spread of unaccountable disinformation. - There needs to be strong law and regulation of the production and release of deepfakes and synthetic media. For Fact-Checkers & Civil Society - Invest in advanced AI - Detection Tools - Co-operate with international fact checking organizations. - Perform digital literacy and public awareness campaigns. For Technology Platforms - Collaborate with supervisory body and civil society to spot issues in time. - Explanation capabilities should be built into AI systems as a policy to enable accountability. - Ensure a proactive monitoring system to ensure that malicious content is caught and stopped before it spreads. Overall, the combination of such strategies is expected to help reduce harmful impacts of Al-generated fake news, but also boost the transparency and accountability and promote the sense of trust towards the national media environment. Importantly, this research represents the first attempt to highlight an integrated framework of ethical, legal and technological framework that directly addresses Al-driven misinformation in the context of Bangladesh's weak democratic and polarized media environment.

Generating HTML Viewer...