Large Language Models (LLMs) and other forms of Generative Artificial Intelligence (GenAI) are growing rapidly, offers extraordinary possibilities in significantly different domains, but this technology also creates new ethical, security, and social dilemmas. In this review, we attempt to synthesize the recent literature on GenAI, including its potential benefits or capabilities, and the critical issues that it raises. There are many articles that covered issues like misinformation, algorithmic bias, privacy, academic integrity, and the “dual use problem”, which we highlight. We also explored current detection and mitigation initiatives, existing governance models, and the need for human-centred, multidisciplinary inquiries. Overall, we found a substantial gap between the speed of aspects of technological development and the progress of appropriate regulations, and we believe that the larger project of building trust and ensuring effective use of GenAI will encompass ongoing innovation and adaptation of our safety measures, policies, and AI literacy among stakeholders.
Paper
Full text
Towards Secure and Ethical Generative AI: Challenges and Perspectives
Semantic Scholar · 2025
Abstract
Large Language Models (LLMs) and other forms of Generative Artificial Intelligence (GenAI) are growing rapidly, offers extraordinary possibilities in significantly different domains, but this technology also creates new ethical, security, and social dilemmas. In this review, we attempt to synthesize the recent literature on GenAI, including its potential benefits or capabilities, and the critical issues that it raises. There are many articles that covered issues like misinformation, algorithmic bias, privacy, academic integrity, and the “dual use problem”, which we highlight. We also explored current detection and mitigation initiatives, existing governance models, and the need for human-centred, multidisciplinary inquiries. Overall, we found a substantial gap between the speed of aspects of technological development and the progress of appropriate regulations, and we believe that the larger project of building trust and ensuring effective use of GenAI will encompass ongoing innovation and adaptation of our safety measures, policies, and AI literacy among stakeholders.