Identification and control of potential occupational safety and health hazards of artificial intelligence: a scoping review

Introduction Artificial intelligence (AI) offers substantial societal benefits, but it also can pose significant hazards to workers. Methods A scoping literature review was conducted to identify materials published from 2010 to 2025 and indicative of artificial intelligence (AI) hazards to workers. The aim was to answer two research questions: What are the potential occupational safety and health hazards associated with exposure to AI? and What are the approaches that have been proposed to control these hazards in the workplace? Results Eight databases and grey literature were searched. These yielded 25,925 studies, which after screening resulted in 622 studies for full text review. Of these, 216 were eligible for data extraction. These studies substantiated 14 categories of potential hazards to workers from AI. These categories included: threats of physical trauma; dehumanization of workers; threats to human agency, autonomy, and workers’ voice; loss of human and workplace connection; threats to sense of self and identity; threats of increased stress, fatigue, and work-life imbalance from work intensification; dependence on AI, leading to deterioration of skills; potential exacerbation of OSH inequities in the workplace; threats of weakening the psychological contract, worker engagement, and trust; threats of job loss, unemployment, disruption of labor market, and reduced economic mobility; threats of underpaid, precarious, and hidden work; threats of cyberphysical attacks and bullying; inappropriate worker monitoring, surveillance, management, and selection; and threats of adverse consequences of worker cognitive and physical augmentation and enhancement. Control approaches also were identified for each hazard category. Discussion This study provides a comprehensive appraisal of AI-related potential OSH hazards and controls and offers a foundation for future research and practice.