Вход на сайт
Просмотр новости

Найдите то, что Вас интересует

Navigating Ethical Complexities in Educational AI: A Systematic Review of Generative Chatbot Integration in Teaching and Learning [version 1; peer review: awaiting peer review]
Дата публикации: 15-07-2026 05:46:17

This systematic literature review examines the ethical challenges associated with integrating generative chatbots in educational contexts. Guided by the PRISMA 2020 framework, the review synthesised peer-reviewed empirical and theoretical studies published between 2022 and 2025 and retrieved from Scopus and Web of Science. Following systematic screening and eligibility assessment, 16 studies were included in the final synthesis. The findings identified four primary ethical domains: student data privacy and protection; academic integrity and AI-assisted plagiarism; algorithmic bias and fairness; and institutional governance and policy readiness. While generative chatbots offer significant pedagogical benefits, including personalised learning, enhanced feedback, and expanded access to educational support, their unregulated use may exacerbate digital inequalities, compromise educational integrity, and reinforce existing biases. The review highlights the need for transparent institutional policies, educator training, ethical AI literacy, strong data governance frameworks, and equitable digital infrastructure. These measures are essential to support the responsible, ethical, and inclusive integration of generative AI technologies in education.

Основное содержимое страницы с новостью.

Introduction

Generative artificial intelligence (AI) chatbots are increasingly reshaping pedagogical practices across educational landscapes. Their capabilities, such as generating real-time feedback, supporting personalised learning pathways, and scaffolding academic writing, have rendered them compelling tools for both educators and learners (Davar et al., 2025; Bayly-Castaneda et al., 2024; AL-Smadi, 2023). In large-scale or resource-constrained educational settings, these tools promise enhanced instructional support and differentiated learning opportunities without proportional increases in human teaching resources (Merino-Campos, 2025).

However, the rapid and often uncritical adoption of generative chatbots in education has surfaced a range of complex ethical challenges. Chief among these are concerns about student data privacy, the potential for AI-facilitated plagiarism, and the reinforcement of structural biases embedded in algorithmic architectures (Li et al., 2023; Yan et al., 2024). These concerns are compounded by disparities in digital infrastructure, varying levels of educator readiness, and the absence of comprehensive institutional policies that guide AI use. For instance, a systematic scoping review of 118 articles highlighted persistent risks of bias, privacy violations, and academic misconduct when large language models are used in educational settings (Yan et al., 2024). Furthermore, research on AI governance in education emphasises the urgent need for institutional safeguards, such as privacy protocols, integrity mechanisms, and transparency frameworks, to ensure the ethical deployment of generative technologies (Al-kfairy et al., 2024).

International policy bodies such as UNESCO (2023) and the OECD (2024) have also stressed the importance of ensuring that AI deployment in education aligns with the principles of equity, accountability, and human-centred design. Their guidelines call for participatory governance, continuous oversight, and context-sensitive implementations that do not exacerbate existing inequalities or compromise pedagogical values. Despite the growing body of research examining generative AI in education, existing evidence remains fragmented across disciplines, educational contexts, and ethical concerns. While individual studies have explored issues such as privacy, academic misconduct, algorithmic bias, and governance, there is a need for a comprehensive synthesis that consolidates current knowledge and identifies common ethical challenges and responses. A systematic review is therefore warranted to provide an integrated understanding of the ethical implications of adopting generative chatbots in education and to inform evidence-based policy and practice.

In response to these developments, the present systematic literature review aims to critically map the ethical terrain surrounding the use of generative chatbots in educational settings. Specifically, the review seeks to synthesise empirical and theoretical research published between 2022 and 2025 to examine four interrelated ethical domains: student privacy and data protection; academic integrity and AI-assisted plagiarism; algorithmic fairness and bias; and institutional governance and policy readiness. The review addresses the following question: What ethical challenges are associated with integrating generative chatbots in educational settings, and what strategies have been proposed to support their responsible, transparent, and equitable implementation? By consolidating diverse perspectives from global and local contexts, this study contributes to the development of evidence-informed strategies for the responsible, transparent, and equitable integration of generative AI technologies in education.

Methodology
Research design

This study employed a systematic literature review (SLR) to critically examine the ethical implications of integrating generative chatbots into educational contexts. The review was guided by the PRISMA 2020 framework (Page et al., 2021), which ensures transparency and rigour in evidence synthesis. The review process followed four structured phases: (1) identification of relevant literature, (2) screening of titles and abstracts, (3) eligibility assessment based on full-text reviews, and (4) final inclusion according to predefined criteria ( Figure 1).

b7b3eb61-b7f6-4279-ae15-ba8c9129cdc4_figure1.gif

Figure 1. PRISMA 2020 flow diagram of study selection.

The figure illustrates the identification, screening, eligibility assessment, and inclusion of studies examining the ethical implications of generative chatbot integration in educational settings. Records were retrieved from Scopus and Web of Science and screened according to.

The primary objective was to synthesise peer-reviewed empirical and theoretical research that addresses ethical concerns across four interrelated domains: student data privacy, academic integrity, algorithmic fairness, and institutional governance in chatbot-supported learning environments. This design enabled the consolidation of interdisciplinary insights to inform evidence-based strategies for the ethical deployment of generative AI in education.

Literature search strategy

To ensure comprehensive selection of literature, searches were conducted across two high-quality academic databases: Scopus and Web of Science, both of which are internationally recognised for indexing peer-reviewed, discipline-relevant scholarship. Searches were conducted in January 2026. Eligibility was restricted to studies published between January 2022 and May 2025 to delimit the review period and capture research emerging during the initial adoption phase of generative AI in education.

A Boolean search strategy was developed to maximise specificity and relevance. Sample search strings included:

(“generative chatbot” OR “AI chatbot” OR “ChatGPT” OR “LLM in education”) AND (ethics OR privacy OR plagiarism OR bias OR “academic integrity” OR “algorithmic fairness”)

Searches were limited to English-language, peer-reviewed journal articles, and no grey literature (e.g., conference abstracts, preprints, white papers, opinion essays) was included, ensuring scholarly rigour and citation reliability.

Inclusion and exclusion criteria

Studies were included if they were peer-reviewed, published between 2022 and 2025, written in English, focused on educational applications of generative AI chatbots or LLMs, and explicitly addressed ethical issues (privacy, bias, plagiarism, governance). Excluded studies included editorials, opinion pieces, non-peer-reviewed sources, non-English publications, studies published before 2022, and studies not situated in educational contexts. For synthesis, included studies were grouped thematically into four domains: student privacy and data protection, academic integrity and AI-assisted plagiarism, algorithmic bias and fairness, and institutional governance and policy readiness. These criteria are summarised in Table 1.

Table 1. Inclusion and exclusion criteria used in the systematic review.CriteriaInclusionExclusionStudy TypeEmpirical studies, theoretical analyses, systematic reviewsEditorials, commentaries, opinion piecesPublication DatePublished between 2022 and 2025Published before 2022LanguageEnglishNon-English Educational ContextFocus on educational deployment (primary, secondary, tertiary) of generative AI toolsNon-educational contexts (e.g., marketing, legal, clinical)Ethical RelevanceExplicit engagement with ethical issues (privacy, bias, plagiarism, governance, etc.)Purely technical studies lacking ethical discussionTechnology TypeStudies discussing generative chatbots or large language models in educationStudies solely on traditional AI or rule-based systems
Screening and selection procedure

All retrieved records were exported to Zotero for reference management and de-duplication. The author conducted title, abstract, and full-text screening using the predefined eligibility criteria. Studies were assessed for methodological transparency, relevance to the ethical domains, and empirical or conceptual contribution to educational discourse. To enhance trustworthiness, the thematic framework and interpretation of findings were reviewed by independent experts in educational technology, digital ethics, and artificial intelligence policy. The study selection process is illustrated in the PRISMA 2020 flow diagram presented in Figure 1.

Following full-text screening, several studies were excluded for failing to meet the predefined eligibility criteria. Although several publications discussed generative artificial intelligence and its educational applications, they were excluded if ethical issues were not the primary focus, educational contexts were insufficiently addressed, or the studies did not provide an explicit analysis of the ethical dimensions of generative chatbots in education. Table 2 summarises the studies excluded after full-text review and provides the specific reasons for exclusion. Reporting these exclusions enhances the transparency and reproducibility of the review process in accordance with PRISMA 2020 reporting recommendations.

Data extraction and analysis

A structured data extraction form was developed to ensure consistency in collecting information across all included studies. This form captured details such as the author(s), year of publication, and source of each study, as well as the educational level and regional context in which the research was conducted. It also recorded the research methodology employed, whether qualitative, quantitative, or mixed-methods, alongside the specific ethical themes addressed and the key findings and practical recommendations presented.

Following data extraction, a thematic synthesis approach was applied to organise the material into four core analytical domains: student privacy and data protection, academic integrity and plagiarism prevention, algorithmic bias and fairness, and institutional governance and policy readiness. This thematic coding process facilitated the identification of cross-cutting ethical issues, enabled comparative analysis across diverse educational and geographical settings, and supported the derivation of evidence-informed strategies to promote the ethical and equitable integration of generative chatbots in education.

Data extraction and thematic analysis

A structured, systematic data extraction procedure was employed to select 16 peer-reviewed studies addressing the ethical deployment of generative chatbots in education. A predesigned extraction template was developed to capture key information on the ethical dimensions of generative AI integration in teaching and learning contexts. Each study was scrutinised for its research objectives, methodological design, ethical focus, educational setting, and primary findings.

Attention was given to the ethical dimensions most commonly explored in these studies, including student data privacy, algorithmic bias, academic integrity (particularly AI-assisted plagiarism), and institutional governance. Relevant methodological features were recorded, such as research approach (qualitative, quantitative, mixed-methods), participant profiles (e.g., educators, students, institutional leaders), data collection tools, and geographical scope. The educational settings of the chatbot implementations were carefully documented, spanning secondary and higher education, as well as various regional contexts. The studies also detailed the nature of chatbot deployment, including whether it was used for automated feedback, tutoring, writing support, or administrative assistance. Particular emphasis was placed on examining reported ethical implications within these implementations.

Following data extraction, a thematic synthesis approach was employed to analyse the data. This process aimed to identify shared ethical concerns, conceptual trends, and recurring recommendations. Through thematic analysis, four principal themes emerged: (1) student privacy and data protection, (2) academic integrity and plagiarism, (3) algorithmic fairness and bias mitigation, and (4) institutional governance, policy, and readiness. These themes serve as the analytical foundation for understanding both the risks and the responses surrounding the adoption of generative chatbots in education. Table 3 presents a PRISMA-aligned summary of the included studies and illustrates the breadth of evidence informing the review. The table highlights the diverse methodological approaches, educational settings, and ethical concerns associated with integrating generative chatbots in education, thereby providing the empirical and conceptual foundation for the thematic analysis that follows.

Table 3. Characteristics and key findings of studies included in the review (n = 16).Authors (Year)Study TypeAI FocusEthical Domain(s)Educational SettingKey Findings Williams (2024)Theoretical analysisChatGPT (generative AI)Data Privacy; Algorithmic Bias; Academic Integrity (plagiarism); Student AutonomyHigher Education (universities, UK context)Generative chatbots promise personalized learning, but raise serious ethics concerns. Handling sensitive student data poses privacy challenges under GDPR/COPPA, and advanced chatbots risk perpetuating societal biases. AI-generated content also threatens academic integrity via plagiarism. Comprehensive measures – clear policies, improved plagiarism detection, new assessment designs – are urged to harness chatbots’ benefits ethically. Cotton, Cotton & Shipway (2023)Conceptual viewpointGPT-3/ChatGPTAcademic Integrity (plagiarism/cheating)Higher Education (global perspective)Early examination of ChatGPT’s impact noted both opportunities (better student engagement and accessibility) and significant risks to honesty and plagiarism. Universities face difficulties detecting AI-assisted dishonesty. The authors suggest institutions develop policies, provide faculty training and student support, and deploy detection methods to ensure AI tools are used ethically and responsibly. Gruenhagen et al. (2024)Empirical survey (n > 300)ChatGPT (assignment help)Academic Integrity (student cheating)Higher Education (University students, Australia)Surveyed students on chatbot use in coursework. A large share admitted using ChatGPT for assignments and did not view it as cheating. This highlights a gap in understanding AI-assisted plagiarism: students are unsure what constitutes misconduct. The study calls for clearer academic integrity guidelines regarding AI, as many students perceive ChatGPT as a legitimate study tool rather than a cheating aid. Evangelista (2025)Systematic literature reviewChatGPT (assessment use)Academic Integrity (exams); Policy Readiness Higher Education (Universities, UAE/global)A comprehensive review of ChatGPT’s impact on assessment found it undermines traditional exams and assignments, requiring urgent changes. The author proposes redesigning exams (e.g. more complex, analytical formats) to be “AI-proof,” deploying advanced AI-detection software, and instituting robust institutional policies on ethical AI use. These strategies aim to preserve academic standards and integrity while still allowing innovative AI use in teaching. Imran & Almusharraf (2023)Systematic review (30 articles)ChatGPT (writing assistant)Academic Integrity (plagiarism); Policy Higher Education (Academic writing, global)This PRISMA review finds ChatGPT offers both opportunities (e.g. improved writing support) and challenges for academic writing. To reap benefits without eroding integrity, academia must update training and policies: instructors should teach students to use AI as a tool (not a crutch) and revise assessment designs and honor codes to address AI-generated work. Policies should clarify acceptable AI use in writing and ensure originality in home exams and assignments. Halaweh (2023)Conceptual analysisChatGPT (general use)Data Privacy; Algorithmic Bias; Academic Integrity Higher Education (General, UAE)One of the first detailed discussions urging responsible AI integration. Educators raised concerns about ChatGPT’s built-in biases and discriminatory outputs, its data privacy issues (user queries may be saved/misused), and plagiarism/cheating risks. The paper argues for embracing ChatGPT in teaching but provides strategies to do so ethically – e.g. using AI outputs as learning aids under strict guidelines so as not to violate academic honesty. Bukar et al. (2024)Systematic review & frameworkChatGPT (policy focus)Academic Integrity; Bias/Fairness; Policy Higher Education (Global policy context)Proposes a “Risk–Reward–Resilience” framework for ChatGPT use in universities. The review (41 studies) shows giving students ChatGPT access boosts productivity (summarizing, etc.) but exposes them to plagiarism and cheating risks. Unlimited information access is a reward, but comes with misinformation and copyright risksDeveloping AI-based plagiarism detectors can strengthen integrity (resilience) but may widen the digital divide and equity gaps. The authors urge policymakers in higher ed to balance these trade-offs with nuanced policies rather than blanket bans. Yan et al. (2024)Systematic scoping reviewLLMs (incl. ChatGPT)Student Privacy; Academic Integrity; Bias; Institutional Governance Higher Education (Global)Discusses misuse, hallucinations, and fairness concerns. Urges robust oversight and responsible adoption frameworks. Pitts, Marcus & Motamedi (2025)Empirical survey (n = 262)AI chatbots (general)Academic Integrity; Accuracy/Bias; Privacy; Policy Higher Education (Undergraduates, USA)A thematic analysis of student perspectives on AI chatbots found the top concern (by far) was academic integrity. Students fear peers using AI to cheat and worry their own honest work might be falsely flagged as AI-generated. Other major concerns include unreliable or hallucinated answers from chatbots and loss of critical-thinking skills due to overreliance. Students also raised data privacy issues and potential AI bias. To address these, the authors urge institutions to establish clear usage policies (what is acceptable AI aid), educate students on verifying AI outputs and maintaining independent skills, and ensure measures for data privacy, bias mitigation, and equitable access to AI tools. Elkhatat (2023)Empirical experimentChatGPT-3.5 vs 4Academic Integrity (plagiarism detection)Higher Education (Written assignments)This study tested whether ChatGPT-generated content can evade plagiarism detection. GPT-3.5 and 4 consistently produced fluent, “original” essays that standard text-matching software struggles to flag. With AI-written work becoming harder to detect, the authors suggest institutions shift focus from purely relying on Turnitin-like tools to cultivating an ethos of integrity: e.g. implementing honor codes and academic integrity pledges. They also advise designing assessments that AI finds difficult (using non-text inputs or oral exams) and teaching students about AI’s knowledge limits (to catch AI’s inaccurate references). Boateng & Boateng (2025)Review & frameworkAI in Ed systems (general AI)Algorithmic Bias & Fairness; Policy Education (Various: admissions, grading, LMS)A broad review focusing on algorithmic bias in educational decision-making found that AI systems can inadvertently reinforce existing inequities. Biases emerge at many stages – from biased training data to opaque algorithms and even in how institutions deploy AI. These biases disproportionately harm marginalized student groups, creating new systemic barriers (e.g. biased admission algorithms affecting racial diversity). The authors propose a comprehensive framework combining technical fixes (fairness metrics, bias mitigation techniques) with policy reforms and transparent institutional guidelines to promote equity. This dual approach (technical + governance) is needed to ensure AI-driven tools in education are fair and accountable. Zhang, Song & Liu (2025)Empirical experimentGenerative AI contentAlgorithmic Bias & Fairness; Privacy School Education (Religious Education context)An experimental study in Scientific Reports examined how generative AI’s built-in biases affect learners in a religious education setting. It found that AI-generated content not only reflects but amplifies cognitive biases, which can skew students’ understanding of diverse religious teachings. While generative AI can personalize learning (e.g. enhance cross-cultural understanding), it also risks reinforcing prejudices, calling it a “double-edged sword”. The authors urge the introduction of ethical guidelines and oversight when deploying generative AI in schools, to ensure inclusive, unbiased educational content and to safeguard values like privacy and autonomy in sensitive contexts. Vartiainen et al. (2025)Empirical study (design-based)Generative AI (text-to-image)Algorithmic Bias (education about bias)Primary/Secondary Education (Finland, 4th & 7th graders)Through hands-on workshops, researchers taught children about AI and algorithmic bias. Over 200 students co-designed simple AI apps and explored biases in AI-generated images. Results showed a significant improvement in children’s understanding of how biased training data can lead to biased outcomes. Students learned to critically evaluate AI technologies after the sessions, they could explain causes of algorithmic bias in their own words and recognized the ethical implications. The study underscores the value of integrating AI ethics and bias awareness into the school curriculum, empowering young learners to be critical and responsible AI users. Golda et al. (2024)Comprehensive surveyGenerative AI (general)Data Privacy & Security Cross-sector (incl. Education)A wide-ranging survey of generative AI privacy/security challenges (covering AI models, applications, attacks, etc.) highlights serious student data privacy issues as AI tools proliferate. The authors stress that safeguarding user data in AI systems requires a multi-faceted approach: developers should adopt “privacy-by-design” principles, institutions must enforce strict data governance and compliance with regulations, and end-users (educators/students) need greater awareness and control over how their data are used. In education, this translates to clearer consent policies, secure AI integrations with learning management systems, and updated laws addressing AI’s data practices. Chan & Hu (2023)Empirical survey (n = 399)AI tools (general)Accuracy & Misinformation; Data Privacy; Ethics Higher Education (University students, Hong Kong)A survey in Hong Kong found students appreciate AI tools’ benefits but have substantial concerns. Chief among these were the accuracy and reliability of AI-generated answers and broader ethical issues. Many worried about misinformation from chatbots and the erosion of academic honesty. Data privacy and security emerged as the students’ most significant concerns as well, alongside fears about AI’s impact on future employment and on human values/skills. The authors suggest institutions provide guidance on verifying AI outputs, address privacy safeguards, and openly discuss the societal implications of AI with students to alleviate these fears. Li et al. (2023)Systematic reviewChatGPT (writing assistant)Academic Integrity; Policy Higher Education (Global)Finds both value and risk in using ChatGPT for writing. Suggests updating policies and integrating AI literacy into curricula.
Validation and reliability

To ensure methodological transparency and reliability, the review followed a structured validation process. The author conducted the title, abstract, and full-text screening using the predefined inclusion and exclusion criteria. Studies were assessed systematically for methodological transparency, relevance to the ethical domains under investigation, and their empirical or conceptual contribution to educational discourse. The consistent application of these criteria helped to enhance transparency and reduce the potential for selection bias throughout the review process.

The methodological quality and thematic interpretation of the included studies were subsequently subjected to expert validation. External experts with expertise in educational technology, digital ethics, and artificial intelligence policy reviewed the thematic framework. Their role was not to determine study eligibility but to evaluate the methodological soundness of the review process, assess the relevance of the identified themes, and verify the alignment between the extracted evidence and the study findings. Feedback from these experts informed refinements to the thematic framework and strengthened the credibility of the analysis.

To further enhance trustworthiness, the analytical framework and thematic interpretations were reviewed by stakeholders with expertise in educational governance, data privacy, and AI literacy. Their critical appraisal provided an additional layer of validation, confirming the conceptual coherence, practical relevance, and applicability of the study findings within contemporary educational contexts.

Reporting and Use of Findings

The final selection of sixteen peer-reviewed studies provides an empirical and conceptual foundation for understanding the ethical implications of integrating generative chatbots in education. These studies, drawn from diverse contexts including the United Kingdom, Australia, the United Arab Emirates, Finland, and Hong Kong, represent a balanced mix of empirical surveys, systematic reviews, policy analyses, and conceptual frameworks. Collectively, they address ethical concerns that map coherently onto four thematic domains: (1) student data privacy and protection, (2) academic integrity and AI-assisted plagiarism, (3) algorithmic bias and fairness, and (4) institutional governance and policy readiness.

The findings were analysed thematically to synthesise recurrent patterns, policy dilemmas, and pedagogical implications. This thematic synthesis is expanded in the discussion section, where the study formulates evidence-based strategies for ethically integrating generative AI tools. The reviewed literature highlights pressing needs for improved data protection measures (Golda et al., 2024; Williams, 2024), frameworks for AI-inclusive academic integrity (Cotton et al., 2023; Elkhatat, 2023), institutional audits of algorithmic fairness (Boateng & Boateng, 2025; Zhang et al., 2025), and governance reforms to keep pace with technological evolution (Bukar et al., 2024; Yan et al., 2024). These findings support the development of comprehensive institutional responses that include teacher training, ethical AI literacy for students, and enforceable usage policies to ensure equitable and responsible AI adoption in diverse educational settings.

Overview

The literature review’s findings are structured into four principal thematic domains based on the ethical challenges most frequently addressed across the selected studies. These are:

  • 1. Student Privacy and Data Protection

  • 2. Academic Integrity and AI-Assisted Plagiarism

  • 3. Algorithmic Bias and Fairness

  • 4. Institutional Governance and Policy Readiness

These themes emerge from both empirical and theoretical studies published between 2022 and 2025, with contributions spanning global contexts and educational levels. Each theme reveals systemic vulnerabilities and areas for intervention, as well as the pedagogical affordances that generative chatbots may enable.

Thematic domain 1: Student privacy and data protection

Concerns about data privacy were highlighted in over two-thirds of the reviewed studies (e.g., Golda et al., 2024; Halaweh, 2023; Chan & Hu, 2023). Eleven studies specifically pointed to the collection, processing, and storage of student-generated data by third-party AI platforms without adequate oversight. Cloud-based generative tools such as ChatGPT pose challenges for compliance with GDPR and POPIA, especially in jurisdictions where legal and institutional frameworks are underdeveloped (Williams, 2024). Additionally, both Evangelista (2025) and Yan et al. (2024) argue that educational institutions are ill-equipped to ensure secure data handling or to offer transparent consent mechanisms. In response, these studies call for implementing institution-specific data governance policies grounded in privacy-by-design principles (Golda et al., 2024).

Thematic domain 2: Academic integrity and AI-assisted plagiarism

Thirteen studies addressed the intersection between generative AI and academic integrity. Across contexts, concerns emerged regarding AI-generated plagiarism, especially in writing-intensive disciplines (Cotton et al., 2023; Imran & Almusharraf, 2023). Experimental work by Elkhatat (2023) demonstrated that standard plagiarism detection tools struggle to identify content generated by ChatGPT-3.5 or 4.0, underscoring the inadequacy of current detection strategies.

Survey studies (Gruenhagen et al., 2024; Pitts et al., 2025) found that students often do not perceive AI-generated assignments as unethical, highlighting a disconnect between institutional policies and students’ understanding. These findings support the development of AI-aware honour codes, assessment designs that evaluate process rather than product, and widespread AI literacy training for both staff and students (Bukar et al., 2024).

Thematic domain 3: Algorithmic bias and fairness

Nine studies focused explicitly on algorithmic bias. Generative chatbots trained on large-scale internet datasets were found to reproduce and amplify societal stereotypes (Boateng & Boateng, 2025; Zhang et al., 2025). These biases are particularly problematic in multicultural or multilingual educational settings, where chatbot outputs may reinforce dominant cultural narratives while marginalising others. Vartiainen et al. (2025) introduced a pedagogical intervention wherein schoolchildren were taught to recognise algorithmic bias through design-based workshops. Such approaches emphasise the importance of integrating AI ethics into school curricula. The broader consensus across studies supports algorithmic transparency, dataset auditing, and collaborative model design involving diverse stakeholders.

Thematic domain 4: Institutional governance and policy readiness

Twelve of the sixteen studies noted that institutional frameworks for AI use in education are either absent or underdeveloped. Despite the increasing use of AI by educators and students, universities and schools often lack coherent strategies for managing ethical risks (Wang et al., 2023; Bukar et al., 2024). The governance vacuum has resulted in inconsistent practices and left educators uncertain about best practices.

Studies such as those by Halaweh (2023) and Pitts et al. (2025) call for agile, inclusive, and transparent policy development. Key recommendations include mandatory teacher training, AI ethics workshops, and institutional codes of conduct addressing both risks and opportunities. Chan & Hu (2023) further emphasise the importance of involving students in co-creating such policies to foster ethical responsibility and buy-in.

Cross-cutting patterns and gaps

The cross-thematic synthesis conducted in this review reveals three principal patterns that cut across the four identified ethical domains. First, there is a notable interdependence of ethical concerns. For example, deficiencies in institutional governance frequently intensify data privacy vulnerabilities and allow algorithmic biases to persist without oversight. Second, marked disparities exist among institutions in terms of preparedness for AI integration. These disparities are often shaped by the availability of digital infrastructure and the extent of faculty development and training initiatives. Third, there is a discernible lack of contextual research. Most of the included studies are situated within higher education institutions in the Global North, with limited representation from the Global South or engagement with primary and secondary education sectors (Zhang et al., 2025; Vartiainen et al., 2025).

This review also identifies several critical gaps in the current body of literature. There is a scarcity of longitudinal research investigating the sustained impact of generative AI on teaching and learning processes. Moreover, participatory research approaches, particularly those involving direct input from students and educators, remain underutilised, thereby limiting the development of user-centred ethical frameworks. Finally, non-Western educational systems are significantly underrepresented, resulting in an incomplete understanding of how generative AI tools interact with diverse pedagogical, cultural, and policy environments. Addressing these gaps is essential for fostering a globally relevant and ethically grounded discourse on AI in education.

Conclusion and Recommendations

This systematic review has illuminated the multifaceted ethical concerns arising from the integration of generative chatbots within educational settings. Drawing upon sixteen peer-reviewed studies published between 2022 and 2025, the review identifies four principal domains of ethical tension: student data privacy and protection, academic integrity and AI-assisted plagiarism, algorithmic bias and fairness, and institutional governance and policy readiness. While these technologies offer considerable pedagogical promise, enhancing personalised feedback, supporting academic writing, and facilitating scalable instruction, their deployment, when left unregulated, may exacerbate educational inequities, erode academic norms, and reinforce systemic discrimination through biased outputs.

The analysis reveals a consistent lack of institutional preparedness across educational sectors, particularly in policy development, ethical oversight, and infrastructure support. In many contexts, students and educators engage with generative AI tools without clear guidelines, appropriate training, or meaningful safeguards. This governance vacuum intensifies risks around data misuse, academic dishonesty, and algorithmic opacity. Furthermore, evidence suggests that many students do not perceive the use of AI-generated content as a breach of academic integrity, highlighting a critical disconnect between institutional expectations and learner perceptions.

To address these challenges, institutions must prioritise developing transparent, context-sensitive policies that govern AI use while safeguarding privacy and intellectual standards. The integration of ethical AI literacy into both student curricula and teacher professional development is essential for cultivating critical awareness, responsible use, and digital fluency. Assessment practices should be redesigned to mitigate the risk of AI-facilitated misconduct, with greater emphasis on process-driven, authentic, and oral or collaborative forms of evaluation.

Equity must be central to the adoption of generative chatbots. This requires institutional investment in digital infrastructure to ensure access for all learners, alongside targeted support for marginalised groups. Developers and educators should collaborate to audit and mitigate algorithmic biases through inclusive design practices and regular scrutiny of training datasets. Data governance must be reimagined to comply with legal and ethical standards, ensuring privacy-by-design, transparency, and user agency in data use.

Finally, future research must extend beyond short-term case studies to include longitudinal and participatory inquiries that reflect the diversity of educational systems, particularly in underrepresented regions. A more global and inclusive research agenda is needed to ensure that the ethical discourse around AI in education remains relevant, equitable, and responsive to varied pedagogical, cultural, and institutional contexts.

The responsible integration of generative chatbots into teaching and learning hinges not only on technical competence but also on ethical clarity, institutional commitment, and shared pedagogical values. By aligning innovation with integrity, educational stakeholders can foster AI-enhanced learning environments that are just, inclusive, and resilient.

Limitations of the review process

This review was limited to English-language peer-reviewed articles indexed in Scopus and Web of Science between 2022 and 2025. Grey literature, conference proceedings, and non-English publications were excluded. Consequently, some relevant evidence may not have been captured.

Registration and protocol

This systematic review was not prospectively registered, and no formal review protocol was published prior to commencement.

Risk of bias statement

Given the heterogeneous nature of the included evidence, which comprised empirical studies, conceptual papers, policy analyses, and systematic reviews, a formal risk-of-bias assessment tool was not considered appropriate. Instead, methodological transparency, relevance to the review objectives, and conceptual contribution were considered during study selection and synthesis.

Data availability

The PRISMA 2020 checklist, PRISMA flow diagram, search strategy, and supplementary review materials supporting this study are publicly available in the Zenodo repository under a Creative Commons Attribution 4.0 International licence.

Repository: Zenodo DOI https://doi.org/10.5281/zenodo.20717152 (mhlongo, thabo., 2026).

Title: PRISMA 2020 Checklist and Supplementary Materials for Ethical Implications of Generative Chatbot Integration in Educational Settings.

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).

References
  •  Adel A, Ahsan A, Davison C: ChatGPT promises and challenges in education: Computational and ethical perspectives. Educ. Sci. 2024; 14(8): 814. Publisher Full Text
  •  Akpan B, Igwe PA, Mpungose CB, et al.: Bibliometric analysis of generative artificial intelligence research in education. arXiv preprint. 2024. Publisher Full Text
  •  Al-kfairy M, Mustafa D, Kshetri N, et al.: A Systematic Review and Analysis of Ethical Challenges of Generative Ai: An Interdisciplinary Perspective. Available at SSRN 4833030. 2024. Publisher Full Text
  •  AL-Smadi M: ChatGPT and Beyond: The Generative AI Revolution in Education. arXiv. 2023. Publisher Full Text
  •  Bayly-Castaneda K, Ramirez-Montoya MS, Morita-Alexander A: Crafting personalized learning paths with AI for lifelong learning: a systematic literature review. Frontiers in Education. Frontiers; 2024, August; Vol. 9. : 1424386. Publisher Full Text
  •  Bettayeb AM, Abu Talib M, Sobhe Altayasinah AZ, et al.: Exploring the impact of ChatGPT: Conversational AI in education. Frontiers in Education. 2024; 9: 1379796. Publisher Full Text
  •  Boateng O, Boateng B: Algorithmic bias in educational systems: Examining the impact of AI-driven decision making in modern education. World Journal of Advanced Research and Reviews. 2025; 25(1): 2012–2017. Publisher Full Text
  •  Bukar UA, Sayeed MS, Razak SFA, et al.: An integrative decision-making framework to guide policies on regulating ChatGPT usage. PeerJ Computer Science. 2024; 10: e1845. PubMed Abstract | Publisher Full Text | Free Full Text
  •  Chan CKY, Hu W: Students’ voices on generative AI: perceptions, benefits, and challenges in higher education. Int. J. Educ. Technol. High. Educ. 2023; 20: 43. Publisher Full Text
  •  Cotton DRE, Cotton PA, Shipway JR: Chatting and cheating: Ensuring academic integrity in the era of ChatGPT. Innov. Educ. Teach. Int. 2023; 61(2): 228–239. Publisher Full Text
  •  Crompton H, Burke D: The educational affordances and challenges of ChatGPT: State of the field. TechTrends. 2024; 68(2): 380–392. Publisher Full Text
  •  Davar NF, Dewan MAA, Zhang X: AI Chatbots in Education: Challenges and Opportunities. Information. 2025; 16(3): 235. Publisher Full Text
  •  Elkhatat AM: Evaluating the authenticity of ChatGPT responses: a study on text-matching capabilities. Int. J. Educ. Integr. 2023; 19: 15. Publisher Full Text
  •  Evangelista EDL: Ensuring academic integrity in the age of ChatGPT: Rethinking exam design, assessment strategies, and ethical AI policies in higher education. Contemporary Educ. Technol. 2025; 17(1): ep559. Publisher Full Text
  •  Fraiwan M, Khasawneh N: A review of ChatGPT applications in education, marketing, software engineering, and healthcare: Benefits, drawbacks, and research directions. arXiv preprint. 2023. Publisher Full Text
  •  Gill SS, Xu M, Patros P, et al.: Transformative effects of ChatGPT on modern education: Emerging era of AI chatbots. arXiv preprint. 2023. Publisher Full Text
  •  Golda A, Mekonen K, Pandey A, et al.: Privacy and security concerns in generative AI: A comprehensive survey. IEEE Access. 2024; 12: 48126. Publisher Full Text
  •  Gruenhagen JH, Sinclair PM, Carroll J-A, et al.: The rapid rise of generative AI and its implications for academic integrity: Students’ perceptions and use of chatbots for assistance with assessments. Computers & Education: Artificial Intelligence. 2024; 7: 100273. Publisher Full Text
  •  Halaweh M: ChatGPT in education: Strategies for responsible implementation. Contemp. Educ. Technol. 2023; 15(2). Publisher Full Text
  •  Imran M, Almusharraf N: Analyzing the role of ChatGPT as a writing assistant at higher education level: A systematic review of the literature. Contemporary Educ. Technol. 2023; 15(4): ep464. Publisher Full Text
  •  Jo H: From concerns to benefits: A comprehensive study of ChatGPT usage in education. Int. J. Educ. Technol. High. Educ. 2024; 21(35): 1–29. Publisher Full Text
  •  Li M, Enkhtur A, Cheng F, et al.: Ethical implications of ChatGPT in higher education: A scoping review. arXiv preprint arXiv:2311.14378. 2023. Publisher Full Text
  •  Merino-Campos C: The impact of artificial intelligence on personalised learning in higher education: A systematic review. Trends in Higher Education. 2025; 4(2): 17. Publisher Full Text
  •  Mohebi L: Empowering learners with ChatGPT: Insights from a systematic literature exploration. Discover Education. 2024; 3(1): 36. Publisher Full Text
  •  OECD: OECD Recommendation on the Ethics of Artificial Intelligence in Education. OECD Publishing; 2024. Reference Source
  •  Page MJ, McKenzie JE, Bossuyt PM, et al.: Updating guidance for reporting systematic reviews: development of the PRISMA 2020 statement. J. Clin. Epidemiol. 2021; 134: 103–112. PubMed Abstract | Publisher Full Text
  •  Pitts G, Marcus V, Motamedi S: Student Perspectives on the Benefits and Risks of AI in Education. arXiv preprint arXiv:2505.02198. 2025. Publisher Full Text
  •  Porayska-Pomsta K, Holmes W, Nemorin S: The ethics of AI in education: Conceptual perspectives on generative systems. arXiv preprint. 2024. Publisher Full Text
  •  UNESCO: Guidelines for the governance of generative AI in education and research. United Nations Educational, Scientific and Cultural Organization; 2023. Reference Source
  •  Vartiainen H, Valtonen T, Kahila J, et al.: ChatGPT and imaginaries of the future of education: insights of Finnish teacher educators. Inf. Learn. Sci. 2025; 126(1/2): 75–90. Publisher Full Text
  •  Wang H, Dang A, Wu Z, et al.: Generative AI in Higher Education: Seeing ChatGPT through universities’ policies, resources, and guidelines [Preprint]. arXiv. 2023. Publisher Full Text
  •  Williams RT: The ethical implications of using generative chatbots in higher education. Frontiers in Education. Frontiers Media SA; 2024, January; Vol. 8. : 1331607. Publisher Full Text
  •  Yan L, Sha L, Zhao L, et al.: Practical and ethical challenges of large language models in education: A systematic scoping review. Br. J. Educ. Technol. 2024; 55(1): 90–112. Publisher Full Text
  •  Zhang J, Song W, Liu Y: Cognitive bias in generative AI influences religious education. Sci. Rep. 2025; 15: 15720. PubMed Abstract | Publisher Full Text | Free Full Text
  •  Mhlongo T: PRISMA 2020 Checklist and Supplementary Materials for Ethical Implications of Generative Chatbot Integration in Educational Settings. [Data set]. Zenodo. 2026. Publisher Full Text

Схожие новости

#Наименование новостиТональностьИнформативностьДата публикации
1Bibliometric analysis of gamification in teacher education during the generative AI era: historical trends, present status, and future trajectories [version 2; peer review: 3 approved, 1 approved with reservations]011.6323-06-2026
2Beyond the Tool: Ecological Considerations for Integrating Generative AI Into Early Childhood Education08.0624-07-2026
3Human–AI Co-Regulation in Adaptive Learning: Developing GPT-Supported Self-Regulated Learning Models [version 1; peer review: awaiting peer review]010.7427-07-2026
4TECHNOLOGY-BASED TEACHING IN HIGHER EDUCATION IN SOUTH AFRICA: NAVIGATING THE POST-COVID LANDSCAPE (A Systematic Review) [version 1; peer review: awaiting peer review]011.216-07-2026
5Conceptual Understanding and Conceptual Change Through Human AI Collaboration in Science Education: An Integrated Conceptual Framework from a Systematic Literature Review [version 1; peer review: awaiting peer review]04.8808-08-2026
6Educational Chronotopes in the Digital Landscape of Higher Education: A Systematic Bibliometric Mapping of Research Trends, 2015–2025 [version 1; peer review: awaiting peer review]06.3406-08-2026
7Этика и академическая честность в эпоху нейросетей: где грань между умным помощником и плагиатом?0801-01-1970
8Singapore: New Transparency Guidelines for Generative AI Chatbots01029-07-2026
9The Attribution-Efficacy Paradox: A Scoping Review of GenAI Interaction and Academic Self-Efficacy in Higher Education0516-07-2026
10STM Report – AI disclosure in research: towards a global reporting standard07.4705-08-2026

Классификация: . Схожих патентов: 0. Схожих новостей: 10. Тональность: 0. Информативность: 10.44. Источник: f1000research.com.