The dark side of generative artificial intelligence: ethical pitfalls, societal risks, and a research agenda for responsible innovation

Generative AI has proliferated rapidly, presenting significant innovative opportunities and simultaneously triggering concerns around privacy, misinformation, bias, and misuse. This study systematically reviewed 2,905 academic publications from the Scopus, Web of Science and IEEE databases from 2023 to Q1 2025, and mapped the research landscape on Generative AI's dark sides by employing bibliometric methods. Key findings reveal four critical risk domains: (1) privacy invasion and data security breaches, (2) algorithmic bias reinforcement and information manipulation, (3) hallucination, misinformation, and deepfakes, and (4) cyber-attack, malware creation, and moral agency. This work highlights the urgent need for proactive governance, inclusive innovation practices, and global cooperation to harness generative AI's potential, while mitigating its risks.