Responsible AI in Academic Research: A Systematic Review of Ethical Principles, Challenges, and Future Directions

Abstract Artificial intelligence (AI) is now present in every stage of the academic research life cycle, yet the rules governing its use remain fragmented. Following PRISMA 2020, this systematic review used 34 structured search strings and citation chasing to search literature published between 2015 and 2025. Of 359 records identified, 124 studies and documents met the inclusion criteria: 40 empirical studies, 43 conceptual analyses or commentaries, 20 policy and guidance documents, 17 reviews and 4 methodological frameworks; 83% were published from 2023 onwards. Guidance converges on a narrow set of standards: AI cannot be an author, material use must be disclosed, and humans remain accountable. Disclosure thresholds, AI in peer review, AI-generated images, AI-text detectors and the regulatory treatment of research remain contested. Challenges fall into three clusters: algorithmic bias; data governance, privacy and consent; and transparency, reproducibility and integrity, where empirical studies document fabricated references, unreliable and biased detectors, and rapidly rising, mostly undisclosed AI use. Emerging responses include lifecycle-based guidance, structured documentation, tailored ethics review and interdisciplinary oversight. We propose an integrated framework linking five principles to five research stages and an agenda emphasizing context-sensitive ethics, public participation and international coordination.