Background: 

The rise of generative artificial intelligence (AI) tools and digital learning platforms has rendered traditional academic integrity mechanisms increasingly inadequate. Universities worldwide are adopting AI-powered systems — including machine learning (ML), natural language processing (NLP), stylometric analysis, and learning analytics — to detect and prevent misconduct. Existing reviews of this literature remain concentrated in studies of detection tools considered in isolation, and the evidence base itself is heavily skewed toward high-income, technologically advanced higher education systems, leaving developing and conflict-affected contexts largely unexamined.

Objective: 

To synthesize peer-reviewed evidence on AI's role in detecting and preventing academic dishonesty in higher education globally, and to critically examine the ethical, institutional, and contextual conditions that shape its effective and equitable deployment.

Methods: 

Six databases were searched systematically (Scopus, Web of Science, ScienceDirect, IEEE Xplore, SpringerLink, Google Scholar) covering 2012–2024. Eligibility was assessed using predefined PICOS criteria. After three-stage screening, 33 peer-reviewed studies were included. Thematic synthesis followed PRISMA 2020 guidelines; risk of bias was assessed for all included studies using the Mixed Methods Appraisal Tool (MMAT), and findings were stratified by study design to indicate evidential strength.

Results: 

Four themes emerged: (1) AI-based detection mechanisms (plagiarism detection, semantic similarity, stylometry, and an emerging body of work on AI-generated text identification, represented by a small subset of the included studies); (2) AI-based prevention strategies (learning analytics, early warning systems, adaptive feedback); (3) ethical and institutional challenges (algorithmic bias, opacity, false positives, data privacy, faculty integration); and (4) critical contextual and evidentiary gaps, with developing and conflict-affected higher education systems — including the Palestinian case — identified as substantially under-represented in the global literature.

Discussion and conclusion: 

AI demonstrates significant potential to enhance academic integrity systems, yet its effectiveness is constrained by algorithmic bias, an evolving and still-emerging generative AI detection literature, and variable institutional readiness. Higher education systems operating under resource constraints or in conflict-affected settings require context-sensitive governance frameworks and targeted capacity-building before meaningful AI integration is feasible, though the present review identifies this as a priority for future empirical research rather than a conclusion supported by direct evidence. AI should complement human judgment within an integrated integrity ecosystem, not replace it. Empirical research in resource-constrained and conflict-affected higher education settings is urgently needed.

Title

Frontiers in Education

Publisher

Frontiers Media SA

Publisher Country

Switzerland

Indexing

Scopus

Impact Factor

2.6

Publication Type

Prtinted only

Volume

11

Year

2026

Pages

14