Abstract
BACKGROUND: Artificial intelligence (AI) is increasingly applied across healthcare and public health, with evidence of benefits including enhanced diagnostics, predictive modelling, operational efficiency, medical education, and disease surveillance.However, potential harms - such as algorithmic bias, unsafe recommendations, misinformation, privacy risks, and sycophantic reinforcement - pose challenges to safe implementation.Far less attention has been directed to the public health threats posed by artificial general intelligence (AGI), a hypothetical form of AI with human-level or greater cognitive capacities. OBJECTIVE: This article explores the benefits and harms of current AI systems, introduces AGI and its distinguishing features, and examines the threats AGI could pose to public health and humanity's survival. DISCUSSION: Unlike 'narrow' AI, AGI could autonomously learn, generalise across domains, and self-improve, potentially achieving superintelligence with unpredictable behaviours.AGI threatens public health through two broad categories: (1) misuse, where adversaries deploy AGI for cyberattacks, disinformation campaigns, or to develop chemical, biological, radiological, and nuclear (CBRN) weapons; and (2) misalignment, where poorly aligned AGI pursues goals in harmful ways, leading to loss of human control, erosion of autonomy, and potentially existential risk.The population-level consequences include widespread unemployment, reduced trust in health systems, catastrophic biological threats, and risks to human survival. CONCLUSION: Healthcare and public health professionals have a critical role in framing AGI risks as health threats, building coalitions akin to historic movements against nuclear war, and collaborating with AI researchers, ethicists, and policymakers.Leveraging their expertise, trust, and global networks, these professionals can help ensure that AI development prioritises human wellbeing and safeguards humanity's future.